Category Archives: Artificial Intelligence

Monitoring and measuring forest ecosystems is a complex challenge because of an existing combination of softwares, collection systems and computing environments that require increasing amounts of energy to power. The University of Maines Wireless Sensor Networks (WiSe-Net) laboratory has developed a novel method of using artificial intelligence and machine learning to make monitoring soil moisture more energy and cost efficient one that could be used to make measuring more efficient across the broad forest ecosystems of Maine and beyond.

Soil moisture is an important variable in forested and agricultural ecosystems alike, particularly under the recent drought conditions of past Maine summers. Despite the robust soil moisture monitoring networks and large, freely available databases, the cost of commercial soil moisture sensors and the power that they use to run can be prohibitive for researchers, foresters, farmers and others tracking the health of the land.

Along with researchers at the University of New Hampshire and University of Vermont, UMaines WiSe-Net designed a wireless sensor network that uses artificial intelligence to learn how to be more power efficient in monitoring soil moisture and processing the data. The research was funded by a grant from the National Science Foundation.

AI can learn from the environment, predict the wireless link quality and incoming solar energy to efficiently use limited energy and make a robust low cost network run longer and more reliably, says Ali Abedi, principal investigator of the recent study and professor of electrical and computer engineering at the University of Maine.

The software learns over time how to make the best use of available network resources, which helps produce power efficient systems at a lower cost for large scale monitoring compared to the existing industry standards.

WiSe-Net also collaborated with Aaron Weiskittel, director of the Center for Research on Sustainable Forests, to ensure that all hardware and software research is informed by the science and tailored to the research needs.

Soil moisture is a primary driver of tree growth, but it changes rapidly, both daily as well as seasonally, Weiskittel says. We have lacked the ability to monitor effectively at scale. Historically, we used expensive sensors that collected at fixed intervals every minute, for example but were not very reliable. A cheaper and more robust sensor with wireless capabilities like this really opens the door for future applications for researchers and practitioners alike.

The study was published Aug. 9, 2022, in the Springers International Journal of Wireless Information Networks.

Although the system designed by the researchers focuses on soil moisture, the same methodology could be extended to other types of sensors, like ambient temperature, snow depth and more, as well as scaling up the networks with more sensor nodes.

Real-time monitoring of different variables requires different sampling rates and power levels. An AI agent can learn these and adjust the data collection and transmission frequency accordingly rather than sampling and sending every single data point, which is not as efficient, Abedi says.

Contact: Sam Schipani, samantha.schipani@maine.edu

See more here:

Artificial intelligence can be used to better monitor Maine's forests, UMaine study finds - UMaine News - University of Maine - University of Maine

The use of artificial intelligence by public bodies is to be monitored by Britains equality regulator for the first time to ensure technologies are not discriminating against people.

There is emerging evidence that bias built into algorithms can lead to less favourable treatment of people with protected characteristics such as race and sex.

The Equality and Human Rights Commission has made tackling discrimination in AI a major strand of its new three-year strategy.

It is today publishing new guidance to help organisations avoid breaches of equality law, including the public sector equality duty (PSED). The guidance gives practical examples of how AI systems may be causing discriminatory outcomes.

From October, the Commission will work with a cross-section of 30 local authorities and other public bodies in England and Scotland to understand how they are using AI to deliver essential services, such as benefits payments, amid concerns that automated systems are inappropriately flagging certain families as a fraud risk.

The EHRC is also exploring how best to use its powers to examine how organisations are using facial recognition technology, following concerns that the software may be disproportionately affecting people from ethnic minorities.

These interventions will improve how organisations use AI and encourage public bodies to take action to address any negative equality and human rights impacts.

Marcial Boo, chief executive of the EHRC, said:

While technology is often a force for good, there is evidence that some innovation, such as the use of artificial intelligence, can perpetuate bias and discrimination if poorly implemented.

Many organisations may not know they could be breaking equality law, and people may not know how AI is used to make decisions about them.

Its vital for organisations to understand these potential biases and to address any equality and human rights impacts.

As part of this, we are monitoring how public bodies use technology to make sure they are meeting their legal responsibilities, in line with our guidance published today. The EHRC is committed to working with partners across sectors to make sure technology benefits everyone, regardless of their background.

The monitoring projects will last several months and will report initial findings early next year.

The Artifical Intelligence in Public Services guidance advises organisations to consider how the PSED applies to automated processes, to be transparent about how the technology is used and to keep systems under constant review.

In the private sector, the EHRC is currently supporting a taxi driver in a race discrimination claim regarding Ubers use of facial recognition technology for identification purposes.

See the article here:

Equality watchdog takes action to address discrimination in use of artificial intelligence - PoliticsHome

Published on: 08/31/2022 15:03 PM

[[VIEW_THIS]]

The Artificial Intelligence (AI) and Emerging Technologies (ET) Partnership Series will hold itsnext meeting,AI/ET Partnership Series #2: AI & Biotech, virtually and in person at the United States Patent and Trademark Office's (USPTO) Silicon Valley Regional Office on September 22, 2022 from 9:30 a.m. to noon PT. During this meeting, panelists from industry and the USPTO will explore various patent policy issues with respect to the biotech industry, including:

A full agenda with speakers will be posted prior to the event. This event is free and open to the public, so register early to attend in person or virtually.

Stay connected with the USPTO by subscribing to regular email updates.

Visit our subscription center at http://www.uspto.gov/subscribe to update or change your email preferences.

This email was sent from an unmonitored mailbox. To contact us, please visit our website http://www.uspto.gov/about/contacts. To ensure that you continue to receive our news and notices, please modify your email filters to allow mail from subscriptioncenter@subscriptions.uspto.gov.

Continue reading here:

Save the date: Artificial Intelligence and Emerging Technologies Partnership meeting #2 on September 22 - United States Patent and Trademark Office

Until recently, coding involved repetitive tasks, and required knowledge of many minute details. These aspects of coding detracted from the truly creative work that developers enjoy, and they slowed developers down.

Now, artificial intelligencetechnology promises to eliminate much of that repetitive work, and developers are no longer thrown off task by having to search the web for those minute details.

The technology works similarly to auto-complete in word processing but writing code instead of plain language and completing whole functions at a time.

Among the latest offerings in AI-powered is Github's Copilot, an AI-powered pair programmer tool available to all developers for $10 a month or $100 per year.

The company claims Copilot can suggest complete methods, boilerplate code, whole unit tests, and even complex algorithms.

With AI-powered coding technology like Copilot, developers can work as before, but with greater speed and satisfaction, so its really easy to introduce, explains Oege De Moor, vice president of GitHub Next. It does help to be explicit in your instructions to the AI.

He explains that during the Copilot technical preview, GitHub heard from users that they were writing better and more precise explanations in code comments because the AI gives them better suggestions.

Users also write more tests because Copilot encourages developers to focus on the creative part of crafting good tests, De Moor explains. So, these users feel they write better code, hand in hand with Copilot.

He adds that it is, of course, important that users are made aware of the limitations of the technology.

Like all code, suggestions from AI assistants like Copilot need to be carefully tested, reviewed, and vetted, he says. We also continuously work to improve the quality of the suggestions made by the AI.

GitHub Copilot is built with Codex -- a descendent of GPT-3 -- which is trained on publicly available source code and natural language.

Because it was trained both on source code and natural language, you can write a comment in English, and then Codex will suggest the code that follows, De Moor explains. In fact, it can even write an entire function or class just given its description in English.

Tabnine CEO Dror Weiss says in the future, AI assistants will be able to review code for developers, create tests automatically, assist with debugging, and do clever automated maintenance operations on systems.

Eventually, every activity that can be automated, will be automated, he says.

From his perspective, a critical feature for organizations is the ability to integrate the specific best practices and code patterns for projects and organizations.

Using this kind of customized AI, organizations will benefit not just from acceleration but also from better consistency and quality of the code, he explains. Another benefit is reducing the time it takes for developers to become highly productive when joining a new project.

One major advantage of AI-assisted coding tools is context-aware code completion.

Microsoft's Visual Studio IntelliCode, for example, is a set of AI-assisted capabilities that enable developers to efficiently complete code with features like argument completion, code formatting, and style rule reference.

IntelliCode is trained on the code of thousands of highly rated open source projects on GitHub, and it uses context from the current code to make relevant recommendations.

Since launching IntelliCode, Microsoft has made updates such as whole-line code completions and refactoring and suggestions that enhance repeated edit experiences to save time for developers.

For organizations planning to implement a strategy involving AI coding assistants, Weiss says making a roadmap is key.

Organizations need to think strategically and have a vision of how they want to leverage AI, even as some essential functionality isn't yet available in any of the products in the market, he says.

He explains a logical first step toward implementing AI assistance would be identifying a specific group of developers and let them use AI based on pre-trained models that learned code patterns from publicly available code.

After a successful implementation, organizations can start rolling out to other groups. In parallel, they can tailor their AI assistance to their needs by creating custom AI models based on their code.

De Moor also points out developers spend much of their time on other tasks, and soon, those other tasks will also benefit from AI assistance.

Examples of these other tasks that are ripe for AI assistance are code review, testing, and refactoring.

Will this change the job of developers? Sure, but for the better, De Moor says. I do not foresee a future where Copilot produces anything useful without human input, but I do see unbridled human creativity, no longer bogged down by irrelevant detail.

He says programming is now about design (decomposing a large problem into smaller ones), and then specifying what the smaller blocks should do -- and the AI will fill in the details.

Weiss adds that as every company is becoming a software company, software development is every organization's most strategic and resource-constrained activity.

Companies are starting to meet the limits of how many developers they can get and getting smaller teams more productive is paramount -- even more so in a downturn as teams could be understaffed, he says. We believe that AI is the most effective way to make developers and teams more productive and will be the natural next step for every organization that has adopted basic DevOps and CI platforms.

In Search of Coding Quality

Modern App Dev: An Enterprise Guide

Can AI Lead the Way in Low Code/No Code App Development?

Read more:

The Power of Artificial Intelligence Coding Assistance - InformationWeek

From Google Home to Alexa, the role of artificial intelligence (AI) seems to have grown over the years. It is now believed that AI will play a greater role when it comes crypto being traded. As a greater number of financial institutions start offering crypto-assets as wealth management offering, the roles of AI-supported trading will become more popular. There are over 4,000 cryptocurrencies and even the oldest coins show large fluctuations in their prices. Likewise, Bitcoin 30-day volatility index is twice the value from 2016 (as per data published on buybitcoinworldwide), Saurav Raaj, founder, director, Wize, a non-fungible token (NFT) infrastructure for businesses company, told FE Digital Currency.

As per industry observers, AI is used in intelligent trading systems for stock market prediction and currency price prediction. As per a report by IEEE Access, Generalised Autoregressive Conditional Heteroskedasticity (GARCH), is a time-series statistical model used for understanding volatility. AI is in the area of market sentiment analysis. Unlike traditional stocks, discussions among trading communities and social media reports, can drive trading decisions. AI with natural language processing (NLP) can analyse market and community sentiments and provide valuable insights to the traders, Raaj added.

Courtesy: IEEE Access, ResearchGate.

It is believed that trading decision is usually based on behavioural biases that cause them to act on an emotion which could lead to mistakes while processing information. AI-guided crypto trading is unlikely to get rid of emotional factors, it is likely to amplify that via machine learning. A deliberate fix in AI programmes to avoid trading at large corrections, and surges may help. Still, it is also likely to slow the usual stop-loss or take-profit exercise, Liquing Yu, Economic Intelligence Units (EIU) analyst on India, Indonesia, and Singapore, said.

Furthermore, industry experts noted that if properly implemented and trained, AI can help eliminate human bias. According to Vikram Pandya, director, Fintech, SP Jain, it definitely helps make scientific decisions backed by data and not by impulse.

According to Business Insider Report in June 2019, there are three areas where AI is used in banking, namely, conversational banking, anti-fraud detection, risk assessment, and credit underwriting. AI-based systems can help to process trading data which can assist traders to make better investment decisions. AI with machine learning (ML) can provide safeguards against such attacks and reduce damages in real-time. In extreme cases, it can be utilised to trigger circuit breakers and even stop trading, added Raaj.

Also Read: From centralisation to decentralisation; how blockchain-oriented fintech can benefit the financial sector

Follow us onTwitter,Facebook,LinkedIn

See the rest here:

From Google Home to Alexa, Artificial Intelligence to play large in trading of cryptocurrencies - The Financial Express

INNISFAIL, Australia (PRWEB) September 02, 2022

Mark C. Giffin announces his entry into the publishing scene with the release of Electronics One: Book 1 (published by Balboa Press AU), a fiction novel that delves into the emerging field of AI, its benefits and disadvantages.

Stephen Frost enjoys becoming a very rich man by being able to use AI. While working hard and inventing an AI doll as well as an electronic body armor and a fuel saving device, he meets his future wife Svetlana who had studied at a facility designed to train women into keeping a millionaire happy and contented. However, Stephen also realized there was a dark side that could come from all this.

Stephens ability in AI almost got Svetlana kidnapped. His own AI attacked the Pentagon, and then, someone else did the same. He knew it would only be a matter of time before someone designed a powerful AI that would be used for evil purposes. His wife, future children, the great American dream, and the safety of people in general were important to him. At the same time, he would want to keep up his research to produce new technologies to help humanity but there would be many challenges ahead.

Artificial intelligence is emerging as more and more real to people, with its pros and cons. The book puts forward possible advantages and threats relating to AI. It is a unique story with Stephen and his wife and artificial intelligence that has been the core of his life, Giffin says. When asked what he wants readers to take away from the book, he answers, Artificial intelligence is an exciting realm with dire consequences if used wrongly. By itself, artificial Intelligence should not go unchecked and out of control. For More details about the book, please visit https://www.balboapress.com/en-au/bookstore/bookdetails/841232-electronics-one

Electronics One: Book 1By Mark C. GiffinSoftcover | 6 x 9in | 200 pages | ISBN 9781982295110E-Book | 200 pages | ISBN 9781982295127Available at Amazon and Barnes & Noble

About the AuthorMark C. Giffin grew up in Queensland, Australia. He worked in the field of medical imaging while maintaining an interest in karate and cars. He became a black belt third dan and still teaches today. His cars include a four-wheel drive and a V8 sports sedan. Over the last 15 years, he has made a number of trips to the U.S. for up to three months, spending most of this time in Florida. He has also had an interest in emerging AI.

Balboa Press Australia is a division of Hay House, Inc., a leading provider in publishing products that specialise in self-help and the mind, body and spirit genre. Through an alliance with the worldwide self-publishing leader Author Solutions, LLC, authors benefit from the leadership of Hay House Publishing and the speed-to-market advantages of the Author Solutions self-publishing model. For more information or to start publishing today, visit balboapress.com.au or call 1-800-844-925.

Share article on social media or email:

Read the original:

New fiction novel delves into the emerging field of artificial intelligence, its benefits and disadvantages - PR Web

DUBLIN--(BUSINESS WIRE)--The "Diagnostic Imaging Market Forecast to 2028 - COVID-19 Impact and Global Analysis by Modality, Application, and End-user" report has been added to ResearchAndMarkets.com's offering.

The global diagnostic imaging market is projected to reach US$ 53,410.59 million by 2028 from US$ 38,034.56 million in 2022.

Rise in prevalence of chronic diseases drives the market growth. Also, the use of Artificial Intelligence (AI) and analytics in diagnostic imaging equipment would act as a future trend in the global diagnostic imaging market.

According to the Centers for Disease Control and Prevention (CDC) report, six in ten Americans live with at least one chronic disease, including heart disease and stroke, cancer, and diabetes. Chronic disease are the leading causes of death and disability in North America and stand as a leading healthcare cost.

According to CDC, the leading chronic diseases accounted for almost US$ 4.1 trillion in annual healthcare costs in America in 2020. Additionally, diagnostic imaging is widely adopted for chronic conditions of the geriatric population as the population is more vulnerable to the above chronic indications. For instance, JMIR Publications revealed that the population aged >60 is expected to rise to 2 billion by 2050 worldwide.

Thus, with the increasing prevalence of aging and chronic diseases, it is essential to focus on healthcare innovation to improve health services. For example, innovation in diagnostic imaging with the support of information and communication technology (ICT) has been used in several settings that assist individuals in diagnosing, treating, and managing chronic diseases better. Also, ICT interventions in diagnostic imaging provide solutions to some of the challenges associated with aging and chronic diseases.

Osteoporosis is a significant health problem globally and is responsible for a severe clinical and financial burden owing to increasing life expectancy. Moreover, osteoporosis increases the chances of falls, fractures, hospitalization, and mortality. The age-standardized prevalence of osteoporosis among the European population is 12% for women and 12.2% for men aged 50-79 years, per the statistics of the National Library of Medicine in 2020.

Therefore, it is mandatory to conduct clinical assessments for early diagnosis and to prevent the onset of complications. Several diagnostic imaging techniques such as computed tomography (CT), magnetic resonance imaging (MRI), and ultrasound imaging provide information on different aspects of the same pathologies for the detection of osteoporosis at an early stage.

For example, MRI provides information on various aspects of bone pathophysiology, and its results play an essential role in diagnosing diseases early in preventing clinical onset and consequences. The factors mentioned above are responsible for driving the overall global diagnostic imaging market.

Artificial intelligence helps improve numerous aspects of the healthcare industry, and diagnostic imaging technique is one of the fields that would benefit greatly. Diagnostic imaging equipment manufacturers worldwide are integrating AI into their products. For example, in September 2018, Nvidia announced launching the Nvidia Clara platform, a combination of software and hardware working together in diagnostic imaging equipment.

Such ground-breaking technology can address the challenges of medical instruments and process enormous amounts of data generated every second that doctors and scientists can easily interpret.

Market Opportunities of Global Diagnostic Imaging Market

Government initiatives that sanction funds and grants for diagnostic imaging services to reach globally are expected to create lucrative opportunities for the overall global diagnostic imaging market growth in the coming years.

The WHO collaborates with partners and manufacturers to develop effective solutions targeting to improve diagnostic services in remote locations. Additionally, the WHO and its partners provide training programs on the use and management of diagnostic imaging, focusing on patient safety.

For example, in February 2022, Siemens Healthineers announced a partnership with UNICEF that assisted in improving access to healthcare in Sub-Saharan Africa for diagnostic techniques.

Key Market Dynamics

Market Drivers

Market Restraints

Market Opportunities

Future Trends

Company Profiles

For more information about this report visit https://www.researchandmarkets.com/r/bjdzvd

The rest is here:

Global Diagnostic Imaging Market Forecast Report 2022: A $53 Billion Market by 2028 - Artificial Intelligence (AI) and Analytics Gaining Traction -...

Introduction

The uvea of the eye is a highly vascular structure including the anterior uvea and the posterior uvea or choroid,1 which are susceptible to breakdown of the blood-aqueous barrier and inflammatory response in cases of various diseases. Uveitis is a common sight-threatening disease that leads to 510% of vision impairment worldwide.2 It has been suggested that there are several markers that can predict the prognosis of the disease, pathogenesis and treatment outcome.3,4 Sauer et al found that elevated levels of interleukin (IL)-1, IL-2, IL-6, interferon (IFN)- and tissue necrosis factor (TNF)- may be implicated in uveitis.5 Additionally, elevated intraocular levels of IL-6 has been associated with idiopathic uveitis and uveitis in Behets disease, sarcoidosis and ankylosing spondylitis.5 For uveal diseases such as uveal melanoma, the most common primary intraocular malignancy in adults,6 limited information is known on the characteristics that predict survivability for patients.8 Ericsson et al established that Human Leukocyte Antigen (HLA)-I expression is upregulated in metastatic disease resulting in a poor prognosis.9

As artificial intelligence (AI) methods are rapidly progressing, breakthrough technologies are changing the landscape of healthcare research with powerful diagnostic and prognostic value.10 Machine learning methods (also referred to as complex AI), supervised and unsupervised, are employed by AI systems to account for complex interaction either by collecting input data including biofluid and tissue to predict output values based on new input samples or by finding underlying patterns in an unlabelled data set to identify sub-cluster and outliers in the data.10

Although AI methods are well described in other healthcare fields, there is limited information on the value of using AI methods in understanding the complex nature of uveal diseases. Machine learning has allowed for more robust discovery of biomarkers that have been approved by the Food and Drug Administration (FDA) to guide treatment which can be valuable in diseases such as uveitis and uveal melanoma.10 Additionally, the biomarkers act as powerful clinical predictors that can individualize treatment options for patients for more desired outcomes.10

Herein, we aim to systematically review the available literature describing the application of AI methods in uveal diseases, highlighting the important biomarkers identified by AI methods for treatment, prognosis, and disease profile. We also characterize the type of AI methods utilized in uveal disease including sample selection and preferred analysis method, goals of the AI, and guide future research in this ever-evolving field.

This systematic review adhered to the Preferred Reporting Items for a Systematic Review and Meta-analysis (PRISMA) guidelines and the protocol was registered in PROSPERO (reg. CRD42020196749).11

The search strategy was developed with the aid of an expert librarian and was conducted across five electronic databases (EMBASE, Medline, Cochrane Central Register of Controlled Trials, Cochrane Database of Systematic Reviews, and Web of Science). The search was initially conducted from inception to August 11, 2020, and updated on August 1, 2021. Terms related to the concepts of ophthalmology and AI/bioinformatics and proteomics, metabolomics, lipidomics were used in the formal search to capture all relevant articles (Supplemental File 1). Additionally, backward and forward citation tracking was conducted for completeness. Gray literature indexes were included via EMBASE.

The inclusion and exclusion criteria were determined prior to screening. The inclusion criteria were as follows: (1) original peer-reviewed studies that analyzed biomarker concentrations to predict or modify patient therapy or outcome/diagnosis in intraocular ophthalmic conditions; (2) studies that analyzed biomarker using any type of AI and/or bioinformatics approaches; (3) articles that studied biomarker samples from vitreous fluid, aqueous fluid, tear fluid, plasma, serum, or ophthalmic biopsies and analyzed a protein, lipid, or metabolite; (4) studies that combined biofluid biomarkers with other types of biomarkers (eg imaging) in their statistical models; and (5) simple regression studies that were longitudinal. The exclusion criteria included (1) articles studying ophthalmic diseases that only affect pediatric patients (eg retinopathy of prematurity), (2) studies on non-human subjects (animal or cell studies), (3) studies utilizing post-mortem samples from eyes, (4) non-English studies, (5) abstracts, or reviews, systemic reviews and meta-analyses. This study is part of a series of review papers focused on use of AI and biofluids, and for this particular study a subset of all studies concerning uveal diseases (uveal melanoma and uveitis) were included.

All studies identified by the databases were imported into Covidence (Covidence, Veritas Health Innovation, Melbourne, Australia) for screening. Upon automatic removal of duplicate articles, the remaining articles underwent two levels of screening: title and abstract and full text by two independent reviewers. Disagreements were resolved at a follow-up consensus meeting mediated by a third reviewer after each level of screening.

A standardized data collection form developed prior to the commencement of data extraction was used to ensure a comprehensive and consistent extraction. Data was extracted by one reviewer followed by a quality check where 10% of the extractions were verified by a second independent reviewer to ensure consistency of extracted data. Key parameters extracted from each article included study population demographics, biofluid biomarker characterization and significance, and the AI/bioinformatics tool used in the analysis.

Data were synthesized for each study including details regarding the biofluid sample, type of analysis conducted, significant biomarkers, and demographic information of mean age and sex. Furthermore, data concerning the type of AI and/or bioinformatic analysis of the biomarkers used in uveal diseases was categorized based on the study objective and utility including disease progression, disease prognosis, disease profile, disease treatment and differentiating between differential diagnosis. Due to the heterogeneity of the study designs and AI methods employed by researchers, a meta-analysis was not undertaken.

The Joanna Briggs Institute Critical Appraisal Tool was used for critical appraisal of the included studies.12 Risk of bias assessment was completed by one independent reviewer, and a quality check of 10% of the articles was completed by a second reviewer to ensure consistency between the data extractors. High ROB was applied to studies that reached up to 49% of questions as yes, moderate ROB was classified as 5069%, and low ROB was classified as greater than 70%.28

The search strategy yielded 27,702 articles from all the databases. After the duplicates were removed, 10,258 studies were screened and a total of 18 studies met the criteria for inclusion in the systematic review. A PRISMA flow-chart summarizing the results of the literature can be found in Figure 1.

Figure 1 PRISMA flow diagram of search strategy.

Abbreviation: AI, artificial intelligence.

The two diseases of interest were uveal melanoma (44%) and uveitis (56%) (Table 1). With regard to study design, 9 studies were cohort studies (50%), 8 are cross-sectional studies (44%) and 1 is a case report (6%). Fifteen studies were conducted retrospectively (83%) and 3 were completed prospectively (17%). The studies were conducted in 9 different countries, with the majority from China (7,39%). The total number of subjects in each study ranged from 18 to 10,453, while the median age of the patients ranged from 30 to 63 years (Table 1).

Table 1 Summary of Study and Patient Characteristics

The most common type of bio-sample taken from the uveal melanoma patients was tissue (63% studies) from of enucleated eyes and aqueous humor in the uveitis patients (50%, Table 2). Other types of biofluid samples were serum, plasma, undifferentiated blood and vitreous humor. The biomarker sample types collected varied across all studies, as 6 studies included cytokines, 6 metabolites, 5 proteins, 2 serum products, 2 at chemokines, 2 at cellular infiltrates, 2 at immune cells, 1 at lipids, 1 at electrolytes, and 1 at stromal cells. Furthermore, the number of individual biomarkers analyzed varied from 1 to 4386 per study with most studies researching less than 10 (50%). Although all except one study found significant biomarkers for their respective study objective, there is little to no overlap in the specific biomarkers found to be significant. The only overlap was that of lactate dehydrogenase (LDH) in 50% of the uveal melanoma studies.8,1719

Ten (56%) studies used machine learning methods, and 13 (72%) studies used regression methods to interpret the data. Of the 10 studies that used machine learning methods, 2 used unsupervised methods, 3 used supervised methods and 5 used a combination of both methods. The studies that used regression analysis all employed supervised methods. The most common complex AI method used was principal component analysis (33%), whereas logistic regression (38%) analysis was the most common regression tool. Other types of complex AI methods used were artificial neuronal network (6%), hierarchal neural network (6%), decision tree analysis (6%), random forest (6%), partial least square-discriminant analysis (25%), and orthogonal projection to latent structure discriminant analysis (6%). In addition to AI methods, there were 8 studies that conducted analysis using bioinformatics. Bioinformatics was used for either pathway analysis (5 out of 8 bioinformatics studies) or cluster analysis (3 out of 8 bioinformatics studies). Most commonly, the studies that utilized bioinformatics in their methodology did so in order to differentiate between disease diagnosis (4 out of 8 bioinformatics studies) and understand disease profile (4 out of 8 bioinformatics studies). Overall the study objectives included disease progression (6%), disease prognosis (50%), disease treatment (28%), disease profile (22%), and differentiating between differential diagnosis (22%).

Of the 10 studies focused on uveitis, 4 focused on disease differentiation in which 3 of the 4 studies used machine learning methods. Curnow et al studied cytokine levels of uveitis-presenting diseases such as Behcet's disease, herpes-induced, Fuchs heterochromic cyclitis and idiopathic uveitis and used cluster analysis and random forest analysis for disease differentiation and specifically found TH 1 cytokines, IL-6, IL-8, CCL2 and IFNy are elevated in idiopathic uveitis.3 Verhagen et al used PCA and PLS-DA to determine that ketoleucine is upregulated in Human-Leukocyte antigen-B27 (HLA-B27) positive acute anterior uveitis, which can be used to differentiate it from HLA-B27 negative acute anterior uveitis.4 Partial least square discriminant analysis (PLS-DA) was also used by Young et al to differentiate between lens-induced uveitis and chronic uveitis, with a sensitivity of 78% and specificity of 85%.13 Additionally, 3 studies used machine learning methods to examine disease profile.1416 Guo et al used PLS-DA to identify 33 potential biomarkers and 10 metabolic pathways related to acute anterior uveitis after conducting metabolic analysis.14 Similarly, Xu et al also used PLS-DA to determine specific amino acids and fatty acids to differentiate between controls and uveitis induced by Vogt-Koyanagi-Harada and Behcets disease.15 Wang et al used PCA to determine the profile of disease for Posner-Schlossman syndrome-induced uveitis and found 14 significant pathways.16 The remaining studies used regression methods to determine treatment outcomes and prognosis.21,24,26,27

Three studies determined factors predictive of treatment outcome; Indini et al used machine learning, whereas Heppt et al, and Nicholas et al used regression methods.1719 Indini et al used unsupervised artificial neural network analysis (ANN) to determine the importance of baseline factors in predicting response to anti-PD1 treatment in a retrospective cohort patient.17 The specific biomarkers found in blood that showed significance in increasing overall survival and response to treatment value were neutrophil-to-lymphocyte ratio (NLR) and baseline lactate dehydrogenase (LDH).17 Similarly, Heppt et al and Nicholas et al found LDH levels as a significant prognostic factor.18,19 Lastly, all studies for UM found biomarkers significant in determining disease prognosis. While most studies employed regression modeling, 3 studies employed complex AI technology. However, each study used a different algorithm modality; Indini et al, as previously stated, used unsupervised ANN analysis, Sun et al used unsupervised hierarchical neural network and Ehlers et al used supervised principal component analysis.7,17,20 Specifically, Sun et al used hierarchical neural network for recognition of BAP1 expression in tissue samples for prognostic utility.21 Additionally, principal component analysis was conducted by Ehlers et al to analyze microarray expression results to determine that Nbs1 is a highly significant prognostic factor that can stand alone.20 There was one study that used bioinformatics to conduct pathway analysis for disease prognosis.21 CTLA-4 was assessed in 33 types of cancers to determine its expression and pathway via KEGG and GO databases by Zhang et al.22

Most of the studies included in this review were of high quality (94%) and 1 was of moderate quality (6%), as highlighted in Figure 2. Of the cohort studies, 56% were unclear in identifying confounding factors and 78% of the studies were unclear or failed to identify strategies to account for the confounding variables. Similarly, 75% of the cohort studies did not describe their strategies for addressing confounding variables. Additionally, all 8 cohort studies (100%) did not clearly define the inclusion criteria for sample selection.

Figure 2 Risk of bias assessment using the Joanna Briggs Institute Critical Appraisal Tool.

Notes: Yes = clearly defined in the study; Unclear = not clearly defined in the study; No = not considered in the study.

To our knowledge, this is the first systematic review that summarises the current advancements of AI for analysis of biomarkers involved in uveal diseases, specifically uveitis and uveal melanoma. Almost all studies found significant biomarkers related to their disease of interest through either regression or machine learning methods, emphasizing the value of AI. However, due to the heterogeneous nature of the biomarkers chosen in each study, no significant biomarkers have been identified consistently across all studies for uveal conditions.

We provided a wide overview of both complex AI methods and regressions models, highlighting their utility. Principal component analysis was used most commonly, in 33% of studies and was found to be a powerful tool to determine significant biomarkers in uveal diseases. Although there is a large variation in types of complex AI used, many showed strong predictive ability. For instance, the value of a random forest analysis was demonstrated by Curnow et al, where with 100% accuracy elevated cytokines were identified in idiopathic uveitis, specifically TH 1 cytokines, IL-6, IL-8, CCL2 and IFNy.3 The results from this study indicate the value of a random forest analysis and its future application in differentiating disease profile of uveitis in Behcet's disease, herpes-induced, and Fuchs heterochromic cyclitis with larger sample sizes.3

Considering that uveal melanoma is one of the most common ocular malignancies with a high risk of developing metastatic cancer, it would be beneficial to determine biomarkers that may predict disease progression, prognosis and treatment outcomes.18 Although the number and type of significant biomarker varied from study to study, there was one biomarker that was found significant across multiple studies. Lactate dehydrogenase (LDH) was found to be an important biomarker for disease prognosis and disease treatment outcome by Indini et al, Lorenzo et al, Heppt et al and Nicholas et al.8,1719 Indini et al determined that elevated baseline serum LDH was negatively correlated with anti-PD1 treatment outcome, whereas Lorenzo et al, Heppt et al and Nicholas et al observed high LDH levels with decreased prognosis.8,1719 LDH has been previously established as an important prognostic biomarker and is incorporated in staging procedures, such as the Padova-Mayo model and AJCC model.19 The ability to use LDH as a validated prognostic marker supports the idea of biomarkers as valuable prognostic tools.29 However, as highlighted by Indini et al, ANN is able to characterize the importance of such biomarkers in reference to treatment outcomes.17 Identification of important biomarkers involved in uveal diseases may enable better diagnostics and guide treatment decisions.19 In the current review, AI methods are used to confirm previous findings and weigh the significance of LDH against other prognostic variables with respect to treatment outcomes.19 Although the number of studies in this review offers a large amount of information regarding significant biomarkers, with a limited number of studies focusing on each biomarker, it is difficult to recognize definitive biomarkers for diagnostic and prognostic application.

Despite the large amount of data provided by the studies in this review, there are limitations that affect the ability to apply this information in a clinical setting. As assessed by the risk of bias, there were no studies that clearly defined the inclusion criteria for the sample, affecting the generalizability of findings and replicability for future studies. Additionally, there was no mention of the reliability of the biomarker sample collection process, which further affects the bias presented in the studies. This could potentially create confounding variables that were failed to be identified. Additionally, limited information is provided on the specificity and sensitivity of the analytic methods used, making it difficult to assess the precise utility of AI methods.

In the current study, we reviewed the literature on the use of AI or bioinformatics to determine significant biomarkers in disease progression, prognosis, differentiation, profile and treatment outcome of uveitis and uveal melanoma. Particularly, using complex AI methods can be used to weigh the merit of significant biomarkers, such as LDH, in order to create staging tools and predict treatment outcome. Identification of these important biomarkers may guide clinicians in clinical decision-making and optimizing management strategies. Although the information presently available has a large degree of heterogeneity, future studies have the potential of creating impactful AI models that can result in clinical tool development and implementation.

The contents of this manuscript may be presented at the International Conference of Ophthalmology (September 9 to September 12, 2022) pending acceptance.

This research was in-part funded by Fighting Blindness Canada.

The authors report no conflicts of interest in this work.

1. Van der Woerdt A. Management of intraocular inflammatory disease. Clin Tech Small Anim Pract. 2001;16(1):5861. doi:10.1053/svms.2001.22807

2. Tsirouki T, Dastiridou A, Symeonidis C, et al. A focus on the epidemiology of uveitis. Ocul Immunol Inflamm. 2018;26(1):216. doi:10.1080/09273948.2016.1196713

3. Curnow SJ, Falciani F, Durrani OM, et al. Multiplex bead immunoassay analysis of aqueous humor reveals distinct cytokine profiles in uveitis. Invest Ophthalmol Vis Sci. 2005;46(11):4251. doi:10.1167/iovs.05-0444

4. Verhagen FH, Stigter ECA, Pras-Raves ML, Radstake TRDJ, de Boer JH, Kuiper JJW. Aqueous humor analysis identifies higher branched chain amino acid metabolism as a marker for human leukocyte Antigen-B27 acute anterior uveitis and disease activity. Am J Ophthalmol. 2019;198:97110. doi:10.1016/j.ajo.2018.10.004

5. Sauer A, Villard O, Creuzot-Garcher C, et al. Intraocular levels of interleukin 17A (IL-17A) and IL-10 as respective determinant markers of toxoplasmosis and viral uveitis. Clin Vaccine Immunol. 2014;22(1):7278. doi:10.1128/cvi.00423-14

6. Krantz BA, Dave N, Komatsubara KM, Marr BP, Carvajal RD. Uveal melanoma: epidemiology, etiology, and treatment of primary disease. Clin Ophthalmol. 2017;11:279289. doi:10.2147/opth.s89591

7. Sun M, Zhou W, Qi X, et al. Prediction of BAP1 expression in uveal melanoma using densely-connected deep classification networks. Cancers. 2019;11(10):1579. doi:10.3390/cancers11101579

8. Lorenzo D, Ochoa M, Piulats JM, et al. Prognostic factors and decision tree for long-term survival in metastatic uveal melanoma. Cancer Res Treatment. 2018;50(4):11301139. doi:10.4143/crt.2017.171

9. Ericsson C, Seregard S, Bartolazzi A, et al. Association of HLA Class I and Class II Antigen expression and mortality in uveal melanoma. Invest Ophthalmol Vis Sci. 2001;42(10):21532156.

10. Yu K-H, Beam AL, Kohane IS. Artificial intelligence in healthcare. Nat Biomed Eng. 2018;2(10):719731. doi:10.1038/s41551-018-0305-z

11. Moher D, Liberati A, Tetzlaff J, Altman DG. Preferred reporting items for systematic reviews and meta-analyses: the PRISMA statement. PLoS Med. 2009;6(7):e1000097. doi:10.1371/journal.pmed.1000097

12. JBI. Critical-appraisal-tools - Critical Appraisal Tools. Joanna Briggs Institute. Jbi.global; 2021. Available from: https://jbi.global/critical-appraisal-tools. Accessed August 18, 2022.

13. Young SP, Nessim M, Falciani F, et al. Metabolomic analysis of human vitreous humor differentiates ocular inflammatory disease; 2009. Available from: http://Www.molvis.org.http://www.molvis.org/molvis/v15/a129/#sec-discussion. Accessed August 18, 2022.

14. Guo J, Yan T, Bi H, et al. Plasma metabonomics study of the patients with acute anterior uveitis based on ultra-performance liquid chromatographymass spectrometry. Graefes Arch Clin Exp Ophthalmol. 2014;252(6):925934. doi:10.1007/s00417-014-2619-1

15. Xu J, Su G, Huang X, et al. Metabolomic analysis of aqueous humor identifies aberrant amino acid and fatty acid metabolism in Vogt-Koyanagi-Harada and Behcets disease. Front Immunol. 2021;12. doi:10.3389/fimmu.2021.587393

16. Wang H, Zhai R, Sun Q, et al. Metabolomic profile of PosnerSchlossman syndrome: a gas chromatography time-of-flight mass spectrometry-based approach using aqueous humor. Front Pharmacol. 2019;10. doi:10.3389/fphar.2019.01322

17. Indini A, Di Guardo L, Cimminiello C, De Braud F, Del Vecchio M. Artificial intelligence estimates the importance of baseline factors in predicting response to Anti-PD1 in metastatic melanoma. Am J Clin Oncol. 2019;42(8):643648. doi:10.1097/coc.0000000000000566

18. Heppt MV, Heinzerling L, Khler KC, et al. Prognostic factors and outcomes in metastatic uveal melanoma treated with programmed cell death-1 or combined PD-1/cytotoxic T-lymphocyte antigen-4 inhibition. Eur J Cancer. 2017;82:5665. doi:10.1016/j.ejca.2017.05.038

19. Nicholas MN, Khoja L, Atenafu EG, et al. Prognostic factors for first-line therapy and overall survival of metastatic uveal melanoma: the Princess Margaret Cancer Centre experience. Melanoma Res. 2018;28(6):571577. doi:10.1097/cmr.0000000000000468

20. Ehlers JP, Harbour JW. NBS1 expression as a prognostic marker in uveal melanoma. Clin Cancer Res. 2005;11(5):18491853. doi:10.1158/1078-0432.ccr-04-2054

21. Sun L, Wu R, Xue Q, Wang F, Lu P. Risk factors of uveitis in ankylosing spondylitis. Medicine. 2016;95(28):e4233. doi:10.1097/MD.0000000000004233

22. Zhang C, Chen J, Song Q, et al. Comprehensive analysis of CTLA-4 in the tumor immune microenvironment of 33 cancer types. Int Immunopharmacol. 2020;85:106633. doi:10.1016/j.intimp.2020.106633

23. Armstrong GW, Lorch AC. A(eye): a Review of Current Applications of Artificial Intelligence and Machine Learning in Ophthalmology. Int Ophthalmol Clin. 2020;60(1):5771. doi:10.1097/iio.0000000000000298

24. Bonacini M, Soriano A, Cimino L, et al. Cytokine profiling in aqueous humor samples from patients with non-infectious uveitis associated with systemic inflammatory diseases. Front Immunol. 2020;11. doi:10.3389/fimmu.2020.00358

25. Johansson CC, Mougiakakos D, Trocme E, et al. Expression and prognostic significance of iNOS in uveal melanoma. Int J Cancer. 2010:NANA. doi:10.1002/ijc.24984

26. Cai J, Qi L, Chen Y, et al. Evaluation of factors for predicting risk of uveitis recurrence in Behcets disease patients. Braz J Med Biol Res. 2020;53(6). doi:10.1590/1414-431x20209118

27. Fabiani C, Vitale A, Rigante D, et al. Predictors of sustained clinical response in patients with Behets disease-related uveitis treated with infliximab and Adalimumab. Clin Rheumatol. 2018;37(6):17151720. doi:10.1007/s10067-018-4092-4

28. Valesan LF, Da-Cas CD, Rus JC, et al. Prevalence of temporomandibular joint disorders: a systematic review and meta-analysis. Clin Oral Investig. 2021;25(2):441453. doi:10.1007/s00784-020-03710-w

29. Palmer SR, Erickson LA, Ichetovkin I, Knauer DJ, Markovic SN. Circulating serologic and molecular biomarkers in malignant melanoma. Mayo Clin Proc. 2011;86(10):981990. doi:10.4065/mcp.2011.0287

Read this article:

The Application of Artificial Intelligence in the Analysis of Biomarke | OPTH - Dove Medical Press

Artificial Intelligence (AI) is transforming our way of life from new forms of social organization and scientific discovery to defense and intelligence. This explosive progress is especially apparent in the subfield of machine learning (ML), where AI systems learn autonomously by identifying patterns in large volumes of data.[1] Indeed, over the last five years, the fields of AI and ML have witnessed stunning advancements in computer vision (e.g., object recognition), speech recognition, and scientific discovery.[2], [3], [4], [5] However, these advances are not without risk as transformative technologies are generally accompanied by a significant risk profile, with notable examples including the discovery of nuclear energy, the Internet, and synthetic biology. Experts are increasingly voicing concerns over AI risk from misuse by state and non-state actors, principally in the areas of cybersecurity and disinformation propagation. However, issues of control for example, how advanced AI decision-making aligns with human goals are not as prominent in the discussion of risk and could ultimately be equally or more dangerous than threats from nefarious actors. Modern ML systems are not programmed (as programming is typically understood), but rather independently developed strategies to complete objectives, which can be mis-specified, learned incorrectly, or executed in unexpected ways. This issue becomes more pronounced as AI becomes more ubiquitous and we become more reliant on AI decision-making. Thus, as AI is increasingly entwined through tightly coupled critical systems, the focus must expand beyond accidents and misuse to the autonomous decision processes themselves.

The principal mid- to long-term risks from AI systems fall into three broad categories: risks of misuse or accidents, structural risks, and misaligned objectives. The misuse or accident category includes things such as AI-enabled cyber-attacks with increased speed and effectiveness or the generation and distribution of disinformation at scale.[6] In critical infrastructures, AI accidents could manifest as system failures with potential secondary and tertiary effects across connected networks. A contemporary example of an AI accident is the New York Stock Exchange (NYSE) Flash Crash of 2010, which drove the market down 600 points in 5 minutes.[7] Such rapid and unexpected operations from algorithmic trading platforms will only increase in destructive potential as systems increase in complexity, interconnectedness, and autonomy.

The structural risks category is concerned with how AI technologies shape the social and geopolitical environment in which they are deployed. Important contemporary examples include the impact of social media content selection algorithms on political polarization or uncertainty in nuclear deterrence and the offense-to-defense balance.[8],[9] For example, the integration of AI into critical systems, including peripheral processes (e.g., command and control, targeting, supply chain, and logistics), can degrade multilateral trust in deterrence.[10] Indeed, increasing autonomy in all links of the national defense chain, from decision support to offensive weapons deployment, compounds the uncertainty already under discussion with autonomous weapons.[11]

Misaligned objectives is another important failure mode. Since ML systems develop independent strategies, a concern is that the AI systems will misinterpret the correct objectives, develop destructive subgoals, or complete them in an unpredictable way. While typically grouped together, it is important to clarify the differences between a system crash and actions executed by a misaligned AI system so that appropriate risk mitigation measures can be evaluated. Understanding the range of potential failures may help in the allocation of resources for research on system robustness, interpretability, or AI alignment.

At its most basic level, AI alignment involves teaching AI systems to accurately capture what we want and complete it in a safe and ethical manner. Misalignment of AI systems poses the highest downside risk of catastrophic failures. While system failures by themselves could be immensely damaging, alignment failures could include unexpected and surprising actions outside the systems intent or window of probability. However, ensuring the safe and accurate interpretation of human objectives is deceptively complex in AI systems. On the surface, this seems straightforward, but the problem is far from obvious with unimaginably complex subtleties that could lead to dangerous consequences.

In contrast with nuclear weapons or cyber threats, where the risks are more obvious, risks from AI misalignment can be less clear. These complexities have led to misinterpretation and confusion with some attributing the concerns to disobedient or malicious AI systems.[12] However, the concerns are not that AI will defy its programming but rather that it will follow the programming exactly and develop novel, unanticipated solutions. In effect, the AI will pursue the objective accurately but may yield an unintended, even harmful, consequence. Googles Alpha Go program, which defeated the world champion Go[13] player in 2016, provides an illustrative example of the potential for unexpected solutions. Trained on millions of games, Alpha Gos neural network learned completely unexpected actions outside of the human frame of reference.[14] As Chris Anderson explains, what took the human brain thousands of years to optimize Googles Alpha Go completed in three years, executing better, almost alien solutions that we hadnt even considered.[15] This novelty illustrates how unpredictable AI systems can be when permitted to develop their own strategies to accomplish a defined objective.

To appreciate how AI systems pose these risks, by default, it is important to understand how and why AI systems pursue objectives. As described, ML is designed not to program distinct instructions but to allow the AI to determine the most efficient means. As learning progresses, the training parameters are adjusted to minimize the difference between the pursued objective and the actual value by incentivizing positive behavior (known as reinforcement learning, or RL).[16],[17] Just as humans pursue positive reinforcement, AI agents are goal-directed entities, designed to pursue objectives, whether the goal aligns with the original intent or not.

Computer science professor Steve Omohundro illustrates a series of innate AI drives that systems will pursue unless explicitly counteracted.[18] According to Omohundro, distinct from programming, AI agents will strive to self-improve, seek to acquire resources, and be self-protective.[19] These innate drives were recently demonstrated experimentally, where AI agents tend to seek power over the environment to achieve objectives most efficiently.[20] Thus, AI agents are naturally incentivized to seek out useful resources to accomplish an objective. This power-seeking behavior was reported by Open AI, where two teams of agents, instructed to play hide-and-seek in a simulated environment, proceeded to horde objects from the competition in what Open AI described as tool use distinct from the actual objective.[21] The AI teams learned that the objects were instrumental in completing the objective.[22] Thus, a significant concern for AI researchers is the undefined instrumental sub-goals that are pursued to complete the final objective. This tendency to instantiate sub-goals is coined the instrumental convergence thesis by Oxford philosopher Nick Bostrom. Bostrom postulated that intermediate sub-goals are likely to be pursued by an intelligent agent to complete the final objective more efficiently.[23] Consider an advanced AI system optimized to ensure adequate power between several cities. The agent could develop a sub-goal of capturing and redirecting bulk power from other locations to ensure power grid stability. Another example is an autonomous weapons system designed to identify targets that develop a unique set of intermediate indicators to determine the identity and location of the enemy. Instrumental sub-goals could be as simple as locking a computer-controlled access door or breaking traffic laws in an autonomous car, or as severe as destabilizing a regional power grid or nuclear power control system. These hypothetical and novel AI decision processes raise troubling questions in the context of conflict or safety of critical systems. The range of possible AI solutions are too large to consider and can only get more consequential as systems become more capable and complex. The effect of AI misalignment could be disastrous if the AI discovers an unanticipated optimal solution to a problem that results in a critical system becoming inoperable or yielding a catastrophic result.

While the control problem is troubling by itself, the integration of multiagent systems could be far more dangerous and could lead to other (as of now unanticipated) failure modes between systems. Just like complex societies, complex agent communities could manifest new capabilities and emergent failure modes unique to the complex system. Indeed, AI failures are unlikely to happen in isolation and the roadmap for multiagent AI environments is currently underway in both the public and private sectors.

Several U.S. government initiatives for next-generation intelligent networks include adaptive learning agents for autonomous processes. The Armys Joint All-Domain Command and Control (JADC2) concept for networked operations and the Resilient and Intelligent Next-Generation Systems (RINGS) program, put forth by the National Institute of Standards and Technology (NIST), are two notable ongoing initiatives.[24], [25] Literature on cognitive Internet of Things (IoT) points to the extent of autonomy planned for self-configuring, adaptive AI communities and societies to steer networks through managing user intent, supervision of autonomy, and control.[26] A recent report from the worlds largest technical professional organization, IEEE, outlines the benefits of deep reinforcement learning (RL) agents for cyber security, proposing that, since RL agents are highly capable of solving complex, dynamic, and especially high-dimensional problems, they are optimal for cyber defense.[27] Researchers propose that RL agents be designed and released autonomously to configure the network, prevent cyber exploits, detect and counter jamming attacks, and offensively target distributed denial-of-service attacks.[28] Other researchers submitted proposals for automated penetration-testing, the ability to self-replicate the RL agents, while others propose cyber-red teaming autonomous agents for cyber-defense.[29], [30], [31]

Considering the host of problems discussed from AI alignment, unexpected side effects, and the issue of control, jumping headfirst into efforts that give AI meaningful control over critical systems (such as the examples described above) without careful consideration of the potential unexpected (or potentially catastrophic) outcomes does not appear to be the appropriate course of action. Proposing the use of one autonomous system in warfare is concerning but releasing millions into critical networks is another matter entirely. Researcher David Manheim explains that multiagent systems are vulnerable to entirely novel risks, such as over-optimization failures, where optimization pressure allows individual agents to circumvent designed limits.[32] As Manheim describes, In many-agent systems, even relatively simple systems can become complex adaptive systems due to agent behavior.[33] At the same time, research demonstrates that multiagent environments lead to greater agent generalization, thus reducing the capability gap that separates human intelligence from machine intelligence.[34] In contrast, some authors present multiagent systems as a viable solution to the control problem, with stable, bounded capabilities, and others note the broad uncertainty and potential for self-adaptation and mutation.[35] Yet, the author admits that there are risks and the multiplicative growth of RL agents could potentially lead to unexpected failures, with the potential for the manifestation of malignant agential behaviors.[36],[37] AI researcher Trent McConaughy highlights the risk from adaptive AI systems, specifically decentralized autonomous organizations (DAO) in blockchain networks. McConaughy suggests that rather than a powerful AI system taking control of resources, as is typically discussed, the situation may be far more subtle where we could simply hand over global resources to self-replicating communities of adaptive AI systems (e.g., Bitcoins increasing energy expenditures that show no sign of slowing).[38]

Advanced AI capabilities in next-generation networks that dynamically reconfigure and reorganize network operations hold undeniable risks to security and stability.[39],[40] A complex landscape of AI agents, designed to autonomously protect critical networks or conduct offensive operations, would invariably need to develop subgoals to manage the diversity of objectives. Thus, whether individual systems or autonomous collectives, the web of potential failures and subtle side-effects could unleash unpredictable dangers leading to catastrophic second- and third-order effects. As AI systems are currently designed, understanding the impact of the subgoals (or even their existence) could be extremely difficult or impossible. The AI examples above illustrate critical infrastructure and national security cases that are currently in discussion, but the reality could be far more complex, unexpected, and dangerous. While most AI researchers expect that safety will develop concurrently with system autonomy and complexity, there is no certainty in this proposition. Indeed, if there is even a minute chance of misalignment in a deployed AI system (or systems) in critical infrastructure or national defense it is important that researchers dedicate a portion of resources to evaluating the risks. Decision makers in government and industry must consider these risks and potential means to mitigate them before generalized AI systems are integrated into critical and national security infrastructure, because to do otherwise could lead to catastrophic failure modes that we may not be able to fully anticipate, endure, or overcome.

Disclaimer: The authors are responsible for the content of this article. The views expressed do not reflect the official policy or position of the National Intelligence University, the National Geospatial Intelligence Agency, the Department of Defense, the Office of the Director of National Intelligence, the U.S. Intelligence Community, or the U.S. Government.

Anderson, Chris. Life. In Possible Minds: Twenty-Five Ways of Looking at AI, by John Brockman, 150. New York: Penguin Books, 2019.

Avatrade Staff. The Flash Crash of 2010. Avatrade. August 26, 2021. https://www.avatrade.com/blog/trading-history/the-flash-crash-of-2010 (accessed August 24, 2022).

Baker, Bowen, et al. Emergent Tool Use From Multi-Agent Autocurricula. arXiv:1909.07528v2, 2020.

Berggren, Viktor, et al. Artificial intelligence in next-generation connected systems. Ericsson. September 2021. https://www.ericsson.com/en/reports-and-papers/white-papers/artificial-intelligence-in-next-generation-connected-systems (accessed May 3, 2022).

Bostrom, Nick. The Superintelligent Will: Motivation and Instrumental Rationality in Advanced Artificial Agents. Minds and Machines 22, no. 2 (2012): 71-85.

Brown, Tom B., et al. Language Models are Few-Shot Learners. arXiv:2005.14165, 2020.

Buchanan, Ben, John Bansemer, Dakota Cary, Jack Lucas, and Micah Musser. Georgetown University Center for Security and Emerging Technology. Automating Cyber Attacks: Hype and Reality. November 2020. https://cset.georgetown.edu/publication/automating-cyber-attacks/.

Byford, Sam. AlphaGos battle with Lee Se-dol is something Ill never forget. The Verge. March 15, 2016. https://www.theverge.com/2016/3/15/11234816/alphago-vs-lee-sedol-go-game-recap (accessed August 19, 2022).

Drexler, K Eric. Reframing Superintelligence: Comprehensive AI Services as General Intelligence. Future of Humanity Institute. 2019. https://www.fhi.ox.ac.uk/wp-content/uploads/Reframing_Superintelligence_FHI-TR-2019-1.1-1.pdf (accessed August 19, 2022).

Duettmann, Allison. WELCOME NEW PLAYERS | Gaming the Future. Foresight Institute. February 14, 2022. https://foresightinstitute.substack.com/p/new-players?s=r (accessed August 19, 2022).

Edison, Bill. Creating an AI red team to protect critical infrastructure. MITRE Corporation. September 2019. https://www.mitre.org/publications/project-stories/creating-an-ai-red-team-to-protect-critical-infrastructure (accessed August 19, 2022).

Etzioni, Oren. No, the Experts Dont Think Superintelligent AI is a Threat to Humanity. MIT Technology Review. September 20, 2016. https://www.technologyreview.com/2016/09/20/70131/no-the-experts-dont-think-superintelligent-ai-is-a-threat-to-humanity/ (accessed August 19, 2022).

Gary, Marcus, Ernest Davis, and Scott Aaronson. A very preliminary analysis of DALL-E 2. arXiv:2204.13807, 2022.

GCN Staff. NSF, NIST, DOD team up on resilient next-gen networking. GCN. April 30, 2021. https://gcn.com/cybersecurity/2021/04/nsf-nist-dod-team-up-on-resilient-next-gen-networking/315337/ (accessed May 1, 2022).

Jumper, John, et al. Highly accurate protein structure prediction with AlphaFold. Nature 596 (August 2021): 583589.

Kallenborn, Zachary. Swords and Shields: Autonomy, AI, and the Offense-Defense Balance. Georgetown Journal of International Affairs. November 22, 2021. https://gjia.georgetown.edu/2021/11/22/swords-and-shields-autonomy-ai-and-the-offense-defense-balance/ (accessed August 19, 2022).

Kegel, Helene. Understanding Gradient Descent in Machine Learning. Medium. November 17, 2021. https://medium.com/mlearning-ai/understanding-gradient-descent-in-machine-learning-f48c211c391a (accessed August 19, 2022).

Krakovna, Victoria. Specification gaming: the flip side of AI ingenuity. Medium. April 11, 2020. https://deepmindsafetyresearch.medium.com/specification-gaming-the-flip-side-of-ai-ingenuity-c85bdb0deeb4 (accessed August 19, 2022).

Littman, Michael L, et al. Gathering Strength, Gathering Storms: The One Hundred Year Study on Artificial Intelligence (AI100) Study Panel Report. Stanford University. September 2021. http://ai100.stanford.edu/2021-report (accessed August 19, 2022).

Manheim, David. Overoptimization Failures and Specification Gaming in Multi-agent Systems. Deep AI. October 16, 2018. https://deepai.org/publication/overoptimization-failures-and-specification-gaming-in-multi-agent-systems (accessed August 19, 2022).

Nguyen, Thanh Thi, and Vijay Janapa Reddi. Deep Reinforcement Learning for Cyber Security. IEEE Transactions on Neural Networks and Learning Systems. IEEE, 2021. 1-17.

Omohundro, Stephen M. The Basic AI Drives. Proceedings of the 2008 conference on Artificial General Intelligence 2008: Proceedings of the First AGI Conference. Amsterdam: IOS Press, 2008. 483492.

Panfili, Martina, Alessandro Giuseppi, Andrea Fiaschetti, Homoud B. Al-Jibreen, Antonio Pietrabissa, and Franchisco Delli Priscoli. A Game-Theoretical Approach to Cyber-Security of Critical Infrastructures Based on Multi-Agent Reinforcement Learning. 2018 26th Mediterranean Conference on Control and Automation (MED). IEEE, 2018. 460-465.

Pico-Valencia, Pablo, and Juan A Holgado-Terriza. Agentification of the Internet of Things: A Systematic Literature Review. International Journal of Distributed Sensor Networks 14, no. 10 (2018).

Pomerleu, Mark. US Army network modernization sets the stage for JADC2. C4ISRNet. February 9, 2022. https://www.c4isrnet.com/it-networks/2022/02/09/us-army-network-modernization-sets-the-stage-for-jadc2/ (accessed August 19, 2022).

Russell, Stewart. Human Compatible: Artificial Intelligence and the Problem of Control. New York: Viking, 2019.

Shah, Rohin. Reframing Superintelligence: Comprehensive AI Services as General Intelligence. AI Alignment Forum. January 8, 2019. https://www.alignmentforum.org/posts/x3fNwSe5aWZb5yXEG/reframing-superintelligence-comprehensive-ai-services-as (accessed August 19, 2022).

Shahar, Avin, and SM Amadae. Autonomy and machine learning at the interface of nuclear weapons, computers and people. In The Impact of Artificial Intelligence on Strategic Stability and Nuclear Risk, by Vincent Boulanin, 105-118. Stockholm: Stockholm International Peace Research Institute, 2019.

Trevino, Marty. Cyber Physical Systems: The Coming Singularity. Prism 8, no. 3 (2019): 4.

Turner, Alexander Matt, Logan Smith, Rohin Shah, Andrew Critch, and Prasad Tadepalli. Optimal Policies Tend to Seek Power. arXiv:1912.01683, 2021: 8-9.

Winder, Phil. Automating Cyber-Security With Reinforcement Learning. Winder.AI. n.d. https://winder.ai/automating-cyber-security-with-reinforcement-learning/ (accessed August 19, 2022).

Zeng, Andy, et al. Socratic Models: Composing Zero-Shot Multimodal Reasoning with Language. arXiv:2204.00598 (arXiv), April 2022.

Zewe, Adam. Does this artificial intelligence think like a human? April 6, 2022. https://news.mit.edu/2022/does-this-artificial-intelligence-think-human-0406 (accessed August 19, 2022).

Zwetsloot, Remco, and Allan Dafoe. Lawfare. Thinking About Risks From AI: Accidents, Misuse and Structure. February 11, 2019. https://www.lawfareblog.com/thinking-about-risks-ai-accidents-misuse-and-structure (accessed August 19, 2022).

[1] (Zewe 2022)

[2] (Littman, et al. 2021)

[3] (Jumper, et al. 2021)

[4] (Brown, et al. 2020)

[5] (Gary, Davis and Aaronson 2022)

[6] (Buchanan, et al. 2020)

[7] (Avatrade Staff 2021)

[8] (Russell 2019, 9-10)

[9] (Zwetsloot and Dafoe 2019)

[12] (Etzioni 2016)

[13] GO is an ancient Chinese strategy board game

[14] (Byford 2016)

[15] (Anderson 2019, 150)

[16] (Kegel 2021)

[17] (Krakovna 2020)

[18] (Omohundro 2008, 483-492)

[19] Ibid., 484.

[20] (Turner, et al. 2021, 8-9)

[21] (Baker, et al. 2020)

[22] Ibid.

[23] (Bostrom 2012, 71-85)

[24] (GCN Staff 2021)

[25] (Pomerleu 2022)

[26] (Berggren, et al. 2021)

[27] (Nguyen and Reddi 2021)

[28] Ibid.

[29] (Edison 2019)

[30] (Panfili, et al. 2018)

[31] (Winder n.d.)

[32] (Manheim 2018)

[33] Ibid.

[34] (Zeng, et al. 2022)

[35] (Drexler 2019, 18)

[36] Ibid.

[37] (Shah 2019)

[38] (Duettmann 2022)

[39] (Trevino 2019)

[40] (Pico-Valencia and Holgado-Terriza 2018)

Continue reading here:

Artificial Intelligence, Critical Systems, and the Control Problem - HS Today - HSToday

A series of free community lectures and film screenings at the University of Wisconsin-Stevens Point will look at what may happen When Robots Rule the World.

Presented by the College of Letters and Science, the series will explore the futuristic portrayal of robots in film, the daily use of artificial intelligence (A.I.) in mundane tasks and the latest advances in the field of human-centered A.I. and its implications.

The series begins Sept. 13 and continues throughout the academic year, featuring lectures by UW-Stevens Point faculty and other experts as well as film screenings and a panel discussion. Events will take place on campus or at the Portage County Public Library and are free and open to the public. The lectures will also be available via live stream on the website, http://www.uwsp.edu/whenrobotsrule.

A lecture, Dare to be Human, kicks off the series at 7 p.m., Tuesday, Sept. 13, at The Encore in the UW-Stevens Point Dreyfus University Center (DUC). Associate Professor Vera Klekovkina, world languages and literatures, will discuss how robots could become pets, friends, confidants, and even romantic partners, and the similarities and differences between robotic and human relationships. Cro Crga Studio will also offer a creative performance.

Additional fall events include:

Human-centered A.I. is an emerging discipline that seeks to empower humans but brings up issues in privacy, equity, security, and transparency.

The series is sponsored by the University Personnel Development Committee Research and Creative Activities Grant.

Read the original post:

UW-Stevens Point to offer series on the future of artificial intelligence - Point/Plover Metro Wire