Category Archives: Ai

The news: A new AI model for summarizing scientific literature can now assist researchers in wading through and identifying the latest cutting-edge papers they want to read. On November 16, the Allen Institute for Artificial Intelligence (AI2) rolled out the model onto its flagship product, Semantic Scholar, an AI-powered scientific paper search engine. It provides a one-sentence tl;dr (too long; didnt read) summary under every computer science paper (for now) when users use the search function or go to an authors page. The work was also accepted to the Empirical Methods for Natural Language Processing conference this week.

AI2

The context: In an era of information overload, using AI to summarize text has been a popular natural-language processing (NLP) problem. There are two general approaches to this task. One is called extractive, which seeks to find a sentence or set of sentences from the text verbatim that captures its essence. The other is called abstractive, which involves generating new sentences. While extractive techniques used to be more popular due to the limitations of NLP systems, advances in natural language generation in recent years have made the abstractive one a whole lot better.

How they did it: AI2s abstractive model uses whats known as a transformera type of neural network architecture first invented in 2017 that has since powered all of the major leaps in NLP, including OpenAIs GPT-3. The researchers first trained the transformer on a generic corpus of text to establish its baseline familiarity with the English language. This process is known as pre-training and is part of what makes transformers so powerful. They then fine-tuned the modelin other words, trained it furtheron the specific task of summarization.

The fine-tuning data: The researchers first created a dataset called SciTldr, which contains roughly 5,400 pairs of scientific papers and corresponding single-sentence summaries. To find these high-quality summaries, they first went hunting for them on OpenReview, a public conference paper submission platform where researchers will often post their own one-sentence synopsis of their paper. This provided a couple thousand pairs. The researchers then hired annotators to summarize more papers by reading and further condensing the synopses that had already been written by peer reviewers.

To supplement these 5,400 pairs even further, the researchers compiled a second dataset of 20,000 pairs of scientific papers and their titles. The researchers intuited that because titles themselves are a form of summary, they would further help the model improve its results. This was confirmed through experimentation.

AI2

Extreme summarization: While many other research efforts have tackled the task of summarization, this one stands out for the level of compression it can achieve. The scientific papers included in the SciTldr dataset average 5,000 words. Their one-sentence summaries average 21. This means each paper is compressed on average to 238 times its size. The next best abstractive method is trained to compress scientific papers by an average of only 36.5 times. During testing, human reviewers also judged the models summaries to be more informative and accurate than previous methods.

Next steps: There are already a number of ways that AI2 is now working to improve their model in the short term, says Daniel Weld, a professor at the University of Washington and manager of the Semantic Scholar research group. For one, they plan to train the model to handle more than just computer science papers. For another, perhaps in part due to the training process, theyve found that the tl;dr summaries sometimes overlap too much with the paper title, diminishing their overall utility. They plan to update the models training process to penalize such overlap so it learns to avoid repetition over time.

In the long-term, the team will also work summarizing multiple documents at a time, which could be useful for researchers entering a new field or perhaps even for policymakers wanting to get quickly up to speed. What we're really excited to do is create personalized research briefings, Weld says, where we can summarize not just one paper, but a set of six recent advances in a particular sub-area.

View original post here:

An AI helps you summarize the latest in AI - MIT Technology Review

By Tessa Venell '08Nov. 19, 2020

Imagine a classroom in the future where teachers are working alongside artificial intelligence partners to ensure no student gets left behind.The AI partners careful monitoring picks up on a student in the back who has been quiet and still for the whole class and the AI partner prompts the teacher to engage the student. When called on, the student asks a question. The teacher clarifies the material that has been presented and every student comes away with a better understanding of the lesson.This is part of a larger vision of future classrooms where human instruction and AI technology interact to improve educational environments and the learning experience.James Pustejovsky, the TJX Feldberg Professor of Computer Science, is working towards that vision with a team led by the University of Colorado Boulder, as part of the new $20 million National Science Foundation-funded AI Institute for Student-AI Teaming.The research will play a critical role in helping ensure the AI agent is a natural partner in the classroom, with language and vision capabilities, allowing it to not only hear what the teacher and each student is saying, but also notice gestures (pointing, shrugs, shaking a head), eye gaze, and facial expressions (student attitudes and emotions).

Pustejovsky took some time to answer questions from BrandeisNOW about his research.

How does your research help build this classroom of the future?For the past five years, we have been working to create a multimodal embodied avatar system, called Diana, that interacts with a human to perform various tasks. She can talk, listen, see, and respond to language and gesture from her human partner, and then perform actions in a 3D simulation environment called VoxWorld. This is work we have been conducting with our collaborators at Colorado State University, led by Ross Beveridge in their vision lab. We are working together again (CSU and Brandeis) to help bring this kind of embodied human computer interaction into the classroom. Nikhil Krishnaswamy, my former Ph.D. student and co-developer of Diana, has joined CSU as part of their team.How does it work in the context of a classroom setting?At first its disembodied, a virtual presence on an iPad, for example, where it is able to recognize the voices of different students. So imagine a classroom: Six to 10 children in grade school. The initial goal in the first year is to have the AI partner passively following the different students, in the way they're talking and interacting, and then eventually the partner will learn to intervene to make sure that everyone is equitably represented and participating in the classroom.Are there other settings that Diana would be useful in besides a classroom?Let's say I've got a Julia Child app on my iPad and I want her to help me make bread. If I start the program on the iPad, the Julia Child avatar would be able to understand my speech. If I have my camera set up, the program allows me to be completely embedded and embodied in a virtual space with her so that she can help me.

Screenshot of the embodied avatar system Diana."

How does she help you?She would look at my table and say, Okay, do you have everything you need. And then Id say, I think so. So the camera will be on, and if you had all your baking materials laid out on your table, she would scan the table. She'd say, I see flour, yeast, salt, and water, but I don't see any utensils: you're going to need a cup, you're going to need a teaspoon. After you had everything you needed, she would tell you to put the flour in that bowl over there. And then she'd show you how to mix it.

Is that where Diana comes in?Yes, Diana is basically becoming an embodied presence in the human-computer interaction: she can see what you're doing, you can see what she's doing. In a classroom interaction, Diana could help with guiding students through lesson plans, through dialogue and gesture, while also monitoring the students progress, mood, and levels of satisfaction or frustration.Does Diana have any uses in virtual learning in education?

Using an AI partner for virtual learning could be a fairly natural interaction. In fact, with a platform such as Zoom, many of the computational issues are actually easier since voice and video tracks of different speakers have already been segmented and identified. Furthermore, in a Hollywood Squares display of all the students, a virtual AI partner may not seem as unnatural, and Diana might more easily integrate with the students online.What stage is the research at now?Within the context of the CU Boulder-led AI Institute, the research has just started. Its a five-year project, and its getting off the ground. This is exciting new research that is starting to answer questions about using our avatar and agent technology with students in the classroom.

The research is funded by the National Science Foundation, and partners with CU Boulder on the research include Brandeis University, Colorado State University, the University of California, Santa Cruz, UC Berkeley, Worcester Polytechnic Institute, Georgia Institute of Technology, University of Illinois at Urbana-Champaign, and University of Wisconsin-Madison.

Read more here:

Artificial intelligence and the classroom of the future | BrandeisNOW - Brandeis University

AI startup RealityEngines.AI changed its name to Abacus.AI in July. At the same time, it announced a $13 million Series A round. Today, only a few months later, it is not changing its name again, but it is announcing a $22 million Series B round, led by Coatue, with Decibel Ventures and Index Partners participating as well. With this, the company, which was co-founded by former AWS and Google exec Bindu Reddy, has now raised a total of $40.3 million.

Abacus co-founder Bindu Reddy, Arvind Sundararajan and Siddartha Naidu. Image Credits: Abacus.AI

In addition to the new funding, Abacus.AI is also launching a new product today, which it calls Abacus.AI Deconstructed. Originally, the idea behind RealityEngines/Abacus.AI was to provide its users with a platform that would simplify building AI models by using AI to automatically train and optimize them. That hasnt changed, but as it turns out, a lot of (potential) customers had already invested into their own workflows for building and training deep learning models but were looking for help in putting them into production and managing them throughout their lifecycle.

One of the big pain points [businesses] had was, look, I have data scientists and I have my models that Ive built in-house. My data scientists have built them on laptops, but I dont know how to push them to production. I dont know how to maintain and keep models in production. I think pretty much every startup now is thinking of that problem, Reddy said.

Image Credits: Abacus.AI

Since Abacus.AI had already built those tools anyway, the company decided to now also break its service down into three parts that users can adapt without relying on the full platform. That means you can now bring your model to the service and have the company host and monitor the model for you, for example. The service will manage the model in production and, for example, monitor for model drift.

Another area Abacus.AI has long focused on is model explainability and de-biasing, so its making that available as a module as well, as well as its real-time machine learning feature store that helps organizations create, store and share their machine learning features and deploy them into production.

As for the funding, Reddy tells me the company didnt really have to raise a new round at this point. After the company announced its first round earlier this year, there was quite a lot of interest from others to also invest. So we decided that we may as well raise the next round because we were seeing adoption, we felt we were ready product-wise. But we didnt have a large enough sales team. And raising a little early made sense to build up the sales team, she said.

Reddy also stressed that unlike some of the companys competitors, Abacus.AI is trying to build a full-stack self-service solution that can essentially compete with the offerings of the big cloud vendors. That and the engineering talent to build it doesnt come cheap.

Image Credits: Abacus.AI

Its no surprise then that Abacus.AI plans to use the new funding to increase its R&D team, but it will also increase its go-to-market team from two to ten in the coming months. While the company is betting on a self-service model and is seeing good traction with small- and medium-sized companies you still need a sales team to work with large enterprises.

Come January, the company also plans to launch support for more languages and more machine vision use cases.

We are proud to be leading the Series B investment in Abacus.AI, because we think that Abacus.AIs unique cloud service now makes state-of-the-art AI easily accessible for organizations of all sizes, including start-ups, Yanda Erlich, a p artner at Coatue Ventures told me. Abacus.AIs end-to-end autonomous AI service powered by their Neural Architecture Search invention helps organizations with no ML expertise easily deploy deep learning systems in production.

See the rest here:

Abacus.AI raises another $22M and launches new AI modules - TechCrunch

A top researcher at the Massachusetts Institute of Technology on Thursday said that artificial intelligence systems developed for medicine must be more transparent and judged against a set of common standards to ensure fairness and equity.

Its really important, whenever we are developing these models, to train them on diverse populations and report their accuracy, slicing and dicing for different subpopulations, Regina Barzilay, a professor of engineering and computer science at MIT, said Thursday at the STAT Summit. Its very easy with these models, if theyre trained on one population and then applied on anotherto provide inequitable care.

Unlock this article by subscribing to STAT+ and enjoy your first 30 days free!

STAT+ is STAT's premium subscription service for in-depth biotech, pharma, policy, and life science coverage and analysis.Our award-winning team covers news on Wall Street, policy developments in Washington, early science breakthroughs and clinical trial results, and health care disruption in Silicon Valley and beyond.

Excerpt from:

A leading AI researcher calls for standards to ensure equity and fairness - STAT

No truck maker has had a more tumultuous 2020 than Nikola Corp. The maker of battery-electric and fuel-cell electric drivetrain components and trucks, went from zero-emissions transport messiah at the end of spring, when the company went public, to beingcompared to Theranosthe fraudulent and now defunct medical device manufacturer once valued at $10 billionshortly after Labor Day.

Thats when a short seller called Hindenburg Research released its now infamousreport levying claimsthat the innovative hydrogen fuel-cell technology purported to make zero-emission long haul possible and inevitable was a sham and that Trevor Milton, Nikolas outspoken firebrand founder and chairman, was a charlatan.

It could not have come at a worse time, as the company had broken ground on its Coolidge, Ariz., factory in August and just days before the Hindenburg Report dropped, had announced a groundbreaking partnership with General Motors, who would supply Nikolas under-development Class 8 semis and pickup trucks with batteries and fuel cells.

Miltonbriefly fought the allegations before resigning. A November regulatory filing revealed the Securities and Exchange Commission and Department of Justice issued subpoenas related to the alleged fraud, which basically accused Nikola of misrepresenting its progress to investors. Nikola denies these claims.

The damage was done, though, as Nikola shares halved from $50 to Sept. 8, when the GM deal was announced, to around $25 currently.

Despite all of these tectonic shifts, Nikola's new CEO Mark Russell portrayed Nikola as still having a foundation solid enough from which to disrupt the clean trucking sector and make a serious impact on climate change when he kicked off the latest ACT Virtual event on Nov. 17.

Nikola CEO Mark RussellPhoto: Nikola Corp.

The companys mission remains the same.

Nikola was founded back in 2015, with a passion and a vision for a future that would be sustainable, that would be clean, and that would leave a better planet for all of those who come after us, Russell said.

Here is the current timeline in which Nikola believes its solutions will be available to aid that sustainability effort:

The CEO pointed to partnerships and investments from Bosch, CNH Industrial and Hanwha totaling $480 million.

Financial analysts agreed that network of partners is a major plus for Nikola.

We think this core criticism is misconceived,JP Morgan said in a statementshortly after the Hindenburg report was released. We believe the ability to execute by leveraging other companies core competences specifically GM, CNH and Bosch is a major positive, not a negative. We also think speed, flexibility and unencumbered readiness to change course quickly with evolving circumstance in a fast-developing market are fundamental and desirable attributes of a disruptor.

If the GM deal, worth $2 billion in Nikola stock, is not approved in December, Russell had previously stated Bosch and Romeo Power would provide the battery and fuel-cell components.

Once the company starts producing trucks next year, initially rolling out the Class 8 Nikola Tre in Ulm, Germany, out of an Iveco plant.

The first batch of prototypes will be completed this year, and will begin validation and testing and regulatory approvals, Russell said. We'll have another batch of those prototypes next year, and then we'll move into serial production next year.

Once that happens, Nikola will have a very large customer base hungry for ZETs.

Carbon emissions from heavy trucking and from transportation in general have grown faster than any other segment of the economy, Russell noted during the ACT keynote. He added that like European nations that have already set forth carbon neutral goals in the next 15 years (with Norway aggressively targeting 2025), the U.S. must get rid of fossil fuels and start moving towards a zero emission future for heavy transportation and commercial trucking.

Many states have already taken broad steps to push the adoption of zero-emission trucks (ZETs), with California leading the way. Through the California Air Resources Board, the state has provided funding for several projects, including battery-electric truck pilots with OEMs Daimler Trucks North America and Volvo Trucks North America, respectively. DTNAs eCascadias, due out at the end of 2021, have accrued half a million miles.BYD also announced DHLwill deploy its Class 8 electric trucks in the Los Angeles area. DHL projects its deployment could remove 300 metric tons of greenhouse gas emissions annually.

Now youve got other states catching up to and even talking about passing California, and other forward-thinking leaders are talking about making these kind of things nationwide, which we support, Russell said.

California and 14 other states, along with Washington D.C., signed theMulti-State Medium- and Heavy-Duty Zero-Emission Vehicle Memorandum of Understandingto incrementally increase the sales of ZETs to 30% in those areas by 2030. And with the Biden-Harris administration poised to take over the executive branch in January, a national push towards electrification is a foregone conclusion. President-elect Joe Bidens infrastructure plan prominently features electric vehicles and charging infrastructure, while Vice President-elect Kamala Harris co-sponsored the Green New Deal.

The Tre and BEV version of the Badger pickup truck will benefit from any incentives and electric infrastructure provided by the government, though it remains unclear how much federal effort and funding hydrogen vehicles will receive in the next four years.

The inaugural crop of electric trucks are limited to a range of under 300 miles, keeping them bound to regional and drayage duty cycles, so for zero-emissions trucking to become a universal solution by the current timetables, hydrogen must also be developed.

Hydrogen has many advantages besides the fact that the zero emission, Russell said. It's so energy dense as a fuel. When batteries are trying to move something heavy far distances, you start to run into the problem of energy density.

The amount of energy in a kilo of hydrogen is the equivalent of about three kilos of fossil fuels, he continued. It's the only thing out there that we know of that can beat fossil fuels as a transportation fuel.

Breakdown of energy density by weight among diesel, battery-electric and fuel cell electric trucks

Recently, most of the traditional heavy-duty OEMs have also explored fuel-cell powertrains.Kenworth and Toyotahave had a partnership in place since 2019 to fit the Kenworth T680 with fuel cells. Toyota Motors North America and Hinowill also begin testing fuel-cell electric trucks (FCET) based off the Hino XL Series on North American roads in 2021.

A fuel cell-powered version of the Hino XL Series is a win-win for both customers and the community. It will be quiet, smooth and powerful while emitting nothing but water, said Tak Yokoo, senior executive engineer with Toyota research and development

Daimler AG and Volvo Groupalso formed a strategic partnership this summer, which became a binding agreement in November.

For us at Daimler Truck AG and our intended partner, the Volvo Group, the hydrogen-based fuel-cell is a key technology for enabling CO2-neutral transportation in the future, said Martin Daum, chairman and member of the board of management at Daimler Truck AG. We are both fully committed to the Paris Climate Agreement for decarbonizing road transport and other areas, and to building a prosperous jointly held company that will deliver large volumes of fuel-cell systems.

Daimler has tried to develop a feasible way to convert liquid hydrogen to electricity since the 1990s. Series production on the first Class 8 FCET, the Mercedes-Benz GenH2 Truck, is slated to begin sometime after 2024.

Navistar has also partnered with Cumminsto develop a fuel-cell optionof the International RH Series.

This vehicle will feature our next generation fuel-cell configuration and provides a springboard for us to advance our hydrogen technology for line haul trucks, said Amy Davis, vice president and president of new power at Cummins.

Hyundai Motor Co. may be the first to market with itsXCIENT Fuel Cell trucksin 2022. Testing should begin next year.

While the questions around electric truck charging are currently being practically figured out and vetted, understanding how to realistically install the hydrogen stations to power fuel-cell electric trucks is still in the nascent stages.

We had to solve the chicken-and-egg problem that's always been involved when you're talking about hydrogen fuel-cell vehicles, Russell said.

They believe they have solved that problem and have 700 hydrogen stations planned in the U.S. and more plans to grow in the European market over the next decade in three phases.

Nikola has also joined a consortium to fuel heavy-duty FCETs that includes Hyundai, Toyota, Air Liquide, Nel and Shell. This will help standardize components such as hoses, nozzles and receptacles, Russell said. The fact that FCETs can fill up in the same time as a diesel truck, versus several hours for EVs, is another major selling point.

Comparison of charging time among diesel, battery-electric and fuel cell electric trucksGraph: Nikola Corp.

Theres one thing that helps Nikola stand apart.

What's unique about Nikola is that we offer a bundled lease, Russell explained. We're going to sell you a fuel-cell truck for your long-range transportation needs, we're going to service and support and maintain that truck for you, and we're going to provide the fuel in one bundled lease.

Russell noted that right now the trucking industry spends $100 billion a year on fuel and maintenance. His predecessor Milton said last Maythat bundling these costs could make a Nikola FCET 20 to 30% less than a diesel truck.

More importantly, Russell stressed, is that the companys overall mission to improve air quality is still on track.

Nikola is about establishing a safe, clean, sustainable, zero emission future for all of us on this planet and helping to solve one of the toughest problems, which is how do you get heavy things, long ways across far distances, Russell concluded.

Read this article:

Do You Have A Vision For AI? - IndustryWeek

Reading scientific papers is a tough job. It might contain language and sections that you might not understand, and not all of it would be interesting to you.

My colleague Tristan is great at traversing through these papers, but Im just a novice. So, I would want a quick summary of a research document to decide if I want to dedicate more time reading it.

Thankfully, researchers at theAllen Institute for Artificial Intelligencehave developed a new model to summarize text from scientific papers, and present it in a few sentences in the form of TL;DR (Too Long Didnt Read).

[Read: Neurals market outlook for artificial intelligence in 2021 and beyond]

The team has rolled this model out to the Allen Institutes Semantic Scholar search engine for papers. Currently, youll only see these TL;DR summaries on papers related to computer science on search results or the authors page.

AI takes the most important parts from the abstract, introduction, and conclusion section of the paper to form the summary.

Researchers first pre-trained the model on the English language. Then they created aSciTLDR data set of over 5,400 summaries of computer science papers. It was further trained on more than 20,000 titles of research papers to reduce dependency on domain knowledge while writing a synopsis.

The trained model was able to summarize documents over 5,000 words in just 21 words on an average thats a compression ratio of 238. The researchers now want to expand this model to papers in fields other than computer science.

You can try out the AI on the Semantic Scholar search engine. Plus, you can read more about summarizing AI in this paper.

Published November 20, 2020 10:35 UTC

The rest is here:

TL;DR: This AI summarizes research papers so you dont have to - The Next Web

In one Telegram group chat about the bot, its owner says that Telegram has blocked mentions of its name. However, WIRED was unable to confirm this or any action taken by Telegram. Neither Telegrams spokesperson or the services founder, Pavel Durov, responded to requests for comment. The company, which is believed to be based in Dubai but has servers all around the world, has never publicly commented about the harm caused by the Telegram bot or its continued position to allow it to operate.

Since it was founded in 2013, Telegram has positioned itself as a private space for free speech, and its end-to-end encrypted mode has been used by journalists and activists around the world to protect privacy and evade censorship. However, the messaging app has run into trouble with problematic content. In July 2017, Telegram said it would create a team of moderators to remove terrorism-related content after Indonesia threatened it with a ban. Apple also temporarily removed it from its App Store in 2018 after finding inappropriate content on the platform.

I think they [Telegram] have a very libertarian perspective towards content moderation and just any sort of governance on their platform, says Mahsa Alimardani, a researcher at the Oxford Internet Institute. Alimardani, who has worked with activists in Iran, points to Telegram notifying its users about a fake version of the app created by authorities in the country. It seems that the times that they have actually acted, it's when state authorities have got involved.

On October 23, Italys data protection body, the Garante per la Protezione dei dati Personali, opened an investigation into Telegram and has asked it to provide data. In a statement, the regulator said the nude images generated by the bot could cause irreparable damage to their victims. Since Italian officials opened their investigation, Patrini has conducted more research looking for deepfake bots on Telegram. He says there are a number of Italian-language bots that appear to offer the same functionalities as the one Sensity previously found, however they do not appear to be working.

Separate research from academics of at the University of Milan and the University of Turin has also found networks of Italian-language Telegram groups, some of which were private and could only be accessed by invitation, sharing non-consensual intimate images of women that dont involve deepfake technology. Some groups they found had more than 30,000 members and required members to share non-consensual images or be removed from the group. One group focused on sharing images of women that were taken in public places without their knowledge.

Telegram should look inward and hold itself accountable, says Honza ervenka, a solicitor at law firm McAllister Olivarius, which specializes in non-consensual images and technology. ervenka says that new laws are needed to force tech companies to better protect their users and clamp down on the use of abusive automation technology. If it continues offering the Telegram Bot API to developers, it should institute an official bot store and certify bots the same way that Apple, Google, and Microsoft do for their app stores. However, ervenka adds there is little government or legal pressure being put in place to make Telegram take this kind of step.

Patrini warns that deepfake technology is quickly advancing, and the Telegram bot is a sign of what is likely to happen in the future. The bot on Telegram was the first time this type of image abuse has been seen at such a large scale, and it is easy for anyone to useno technical expertise is needed. It was also one of the first times that members of the public were targeted with deepfake technology. Previously celebrities and public figures were the targets of non-consensual AI porn. But as the technology is increasingly democratized, more instances of this type of abuse will be discovered online, he says.

This was one investigation, but we are finding these sorts of abuses in multiple places on the internet, Patrini explains. There are, at a smaller scale, many other places online where images are stolen or leaked and are repurposed, modified, recreated, and synthesized, or used for training AI algorithms to create images that use our faces without us knowing.

This story originally appeared on WIRED UK.

More Great WIRED Stories

Read the original here:

Telegram Still Hasn't Removed an AI Bot That's Abusing Women - WIRED

On November 17, 2020, the Office of Management Budget (OMB) issued Memorandum M-21-06, Guidance for Regulation of Artificial Intelligence Applications providing policy considerations for AI applications that are developed and deployed by the private sector.

The Memorandum notes that fostering innovation and growth in AI is the Governments top priority and thus agencies must avoid implementing policies or regulations that hamper these priorities. To that end, the Memorandum: (A) propounds ten principles that agencies should consider when determining whether regulation over AI capabilities are necessary; (B) identifies four non-regulatory actions that agencies can take outside the rulemaking process; and (C) addresses how to reduce barriers to the deployment and use of AI.

The ten principles that agencies should consider when formulating regulatory and non-regulatory AI approaches are: (1) Public Trust in AI; (2) Public Participation, including from industry; (3) Scientific Integrity and Information Quality; (4) Risk Assessment and Management; (5) Benefits and Costs; (6) Flexibility; (7) Fairness and Non-Discrimination; (8) Disclosure and Transparency; (9) Safety and Security; and (10) Interagency Coordination.

The Memorandum notes that agencies may find that regulation is unnecessary but that a non-regulatory approach may be warranted to address the risks of AI applications. The Memorandum provides examples of such approaches, including: (1) Sector-Specific Policy Guidance or Frameworks; (2) Pilot Programs and Experiments, including periodically informing industry and the public about emerging trends and changes; (3) Voluntary Consensus Standards, including involving the private sector and other stakeholders in the development of such standards; and (4) Voluntary Frameworks.

The Memorandum also addresses how to reduce barriers to the deployment and use of AI. The memo states that agencies can: (1) provide access to federal data and models for AI R&D; (2) communicate with the public about the benefits and risks of utilizing AI; (3) engage with the private sector regarding the development of voluntary consensus standards; and (4) engage in international regulatory cooperation.

Finally, the Memorandum requires agencies to submit a plan that identifies any regulations that govern the agencys use of AI applications and also to list and describe the agencys plan to regulate AI. Agency plans are due to OMB by May 17, 2021.

The key takeaway for federal government contractors is that the Government is making a big push with respect to AI governance and safety, and now is the time for industry to get involved through opportunities such as public-private working groups and notice and comment rulemaking. Doing so can help shape any regulatory or non-regulatory actions that may soon apply to industrys development and deployment of AI applications.

Read the rest here:

The Government's Momentum in Ensuring Dominance in AI Continues as OMB Issues Guidance for Regulation of AI Applications - Lexology

Vosloo led the drafting of a new set of guidelines from Unicef designed to help governments and companies develop AI policies that consider childrens needs. Released on September 16, the nine new guidelines are the culmination of several consultations held with policymakers, child development researchers, AI practitioners, and kids around the world. They also take into consideration the UN Convention on the Rights of the Child, a human rights treaty ratified in 1989.

The guidelines arent meant to be yet another set of AI principles, many of which already say the same things. In January of this year, a Harvard Berkman Klein Center review of 36 of the most prominent documents guiding national and company AI strategies found eight common themesamong them privacy, safety, fairness, and explainability.

Rather, the Unicef guidelines are meant to complement these existing themes and tailor them to children. For example, AI systems shouldnt just be explainablethey should be explainable to kids. They should also consider childrens unique developmental needs. Children have additional rights to adults, Vosloo says. Theyre also estimated to account for at least one-third of online users. Were not talking about a minority group here, he points out.

In addition to mitigating AI harms, the goal of the principles is to encourage the development of AI systems that could improve childrens growth and well-being. If theyre designed well, for example, AI-based learning tools have been shown to improve childrens critical-thinking and problem-solving skills, and they can be useful for kids with learning disabilities. Emotional AI assistants, though relatively nascent, could provide mental-health support and have been demonstrated to improve the social skills of autistic children. Face recognition, used with careful limitations, could help identify children whove been kidnapped or trafficked.

Children should also be educated about AI and encouraged to participate in its development. It isnt just about protecting them, Vosloo says. Its about empowering them and giving them the agency to shape their future.

Talking about disadvantaged groups, of course children are the most disadvantaged ones.

Unicef isnt the only one thinking about the issue. The day before those draft guidelines came out, the Beijing Academy of Artificial Intelligence (BAAI), an organization backed by the Chinese Ministry of Science and Technology and the Beijing municipal government, released a set of AI principles for children too.

The announcement follows a year after BAAI released the Beijing AI principles, understood to be the guiding values for Chinas national AI development. The new principles outlined specifically for children are meant to be a concrete implementation of the more general ones, says Yi Zeng, the director of the AI Ethics and Sustainable Development Research Center at BAAI who led their drafting. They closely align with Unicefs guidelines, also touching on privacy, fairness, explainability, and child well-being, though some of the details are more specific to Chinas concerns. A guideline to improve childrens physical health, for example, includes using AI to help tackle environmental pollution.

While the two efforts are not formally related, the timing is also not coincidental. After a flood of AI principles in the last few years, both lead drafters say creating more tailored guidelines for children was a logical next step. Talking about disadvantaged groups, of course children are the most disadvantaged ones, Zeng says. This is why we really need [to give] special care to this group of people. The teams conferred with one another as they drafted their respective documents. When Unicef held a consultation workshop in East Asia, Zeng attended as a speaker.

Unicef now plans to run a series of pilot programs with various partner countries to observe how practical and effective their guidelines are in different contexts. BAAI has formed a working group with representatives from some of the largest companies driving the countrys national AI strategy, including education technology company TAL, consumer electronics company Xiaomi, computer vision company Megvii, and internet giant Baidu. The hope is to get them to start heeding the principles in their products and influence other companies and organizations to do the same.

Both Vosloo and Zeng hope that by articulating the unique concerns AI poses for children, the guidelines will raise awareness of these issues. We come into this with eyes wide open, Vosloo says. We understand this is kind of new territory for many governments and companies. So if over time we see more examples of children being included in the AI or policy development cycle, more care around how their data is collected and analyzedif we see AI made more explainable to children or to their caregiversthat would be a win for us.

View post:

Why kids need special protection from AIs influence - MIT Technology Review

The amped-up efforts by Facebook and to tone down blatant misinformation on the campaign trail merits public support. Personally, I find myself trying with limited success to tune out the political noise while sensing that the problem goes beyond that.

The rhetoric of politics overall sounds tired and anachronistic, but then, to my ear, so does much of the dialogue on the popular streamers we binge on. Further, check out the virtual learning classes that now pass for education and you run into even lazier forms of communication. We all decided the earth was flat even before the new Netflix documentary, titled Social Dilemma, pointed up the random anti-truths directed our way.

So while misinformation is being challenged by the social media monoliths, my techno-nerd friends remind me that the demise of honest communication demands a more drastic approach. Their solution? Get ready to groan remember, theyre nerds.

Related StoryPeter Bart: Reed Hastings' Memoir Reveals The Hollywood Re-Education Of A Techie

Their solution is to alert us to the expanding tools of neuro-symbiotic AI artificial intelligence. For most of us, AI conjures an old Steven Spielberg movie in which a robotic Haley Joel Osment keeps flunking the tests of his cybertronics instructors. Little wonder the poor kid kept saying I see dead people (oops, different movie).

But a San Francisco-based software company called Open AI last month unveiled a system that could write coherent essays, design software applications and even propose recipes for breakfast burritos that is, if fed the appropriate maze of symbols. Its called deep learning but it could even lead to deep communicating.

Mike Davies, director of Intel Corps neuromorphic computing lab, contends that neuro-symbolic AI can potentially deliver our own voice assistants adjusted to user needs, analyzing problems or even, some day, writing film scripts or political speeches.

These systems are still nascent but you could imagine that as the technology progresses, entirely new fields could emerge in terms of advertising or media, Francesco Marconi, founder of Applied XI, told the Wall Street Journal. His company generates briefs on health and environmental data. They will become effective at assisting people because theyll be able to understand and communicate.

The ultimate aim is to build support for a sort of Manhattan Project, akin to the body that fostered the atom bomb. Spending on this technology could grow to $3.2 billion by 2023, according to IDC, a research firm, that looks for future support coming from health care, banking and retail. Yann LeCun, chief AI scientist at Facebook, insists we are in sight of creating a machine that can learn how the world works by watching video, listening to audio and reading text.

Given the critical results of its own self-audits, Facebook is under growing pressure to police hate speech, with AI-based censors potentially mobilized to crack down on targeted content. Thus extremists who argue that conservationists triggered the fires in Oregon could no longer aim their social media propaganda directly at any user who happens to check on fires or conservation.

But advances must come from sources even more esoteric then AI, some scientists insist. In a new book titled Livewired, David Eagleman, a practicing futurist, argues that the increasingly important field of brain science itself will nurture development of artificial neural networks.

As these networks proliferate, they will be embellished by machines that themselves can learn, and adjust to new surroundings, such as self-driving cars or power grids distributing electricity.

Argues Eagleman, The capacity to grow new neural circuits isnt just a response to trauma its with all of us every day and it forms the basis of all learning.

So heres the epiphany: Given the heightened sophistication of our neuro circuitry, political candidates may actually have to talk honestly to us. And there is nothing more intimidating to a political candidate than an intelligent audience even if its artificially intelligent.

Original post:

Peter Bart: Does Social Media Misinformation Anger You? Theres An AI For That Too - Deadline