Daily Archives: April 22, 2022

Artificial intelligence was thrust into the public consciousness in the late 1960s courtesy of the Stanley Kubrick movie, "2001: A Space Odyssey." Now our fascination with AI has broadened to include its potential to revolutionize the business world. We're able to spot changing trends and consumer habits in near real time, automatically detect fraud and anomalous behavior, and help sales and marketing teams fine-tune messaging to hit targeted audiences with laser precision.

No wonder 81 percent of senior executives say their business uses AI (up from 48 percent in 2018), and 93 percent report that AI makes their business more competitive, according to the Emerging Tech Executive Report from RELX.

But all is not well in the land of AI. Big HR challenges lie firmly across the path of broad adoption and the realization of AI's potential.

A talent shortage stands in the way of the deployment and full utilization of AI in U.S. companies. Whether through lack of training, difficulties in hiring or headhunters luring away top talent, AI technologists are in short supply. Yet they are needed more than ever, and their skills will be coveted for many years to come.

Supply and Demand

Another study by application development firm Reign, based in Santiago, Chile, found that the U.S. witnessed a 21 percent increase in the number of AI jobs, plus a 27 percent increase in AI-related job wages, over the past decade.

"The demand for workers who can develop AI technology is increasing, as are the effects of AI on workers around the world," said Felipe Silberstein, vice president of strategy at Reign. "Jobs requesting AI or machine-learning skills are expected to increase by 71 percent in the next five years."

The demand is there. So what is choking the AI market and its anticipated expansion? The COVID-19 pandemic can perhaps be partially blamed for slowing the momentum of AI. Yet the RELX survey revealed that the pandemic spurred 48 percent of executives to invest in new AI technologies. The Great Resignation, then, is probably more at fault. With millions quitting their jobs during 2021, companies have been left short-staffed and scrambling for talent. A shortage of resources prevents some companies from maximizing their AI capabilities.

"Companies have scaled up their AI capabilities and are hiring more technologists who are AI-savvy," said Vijay Raghavan, RELX technology forum director. "At the same time, American workers are reconsidering the role that work plays in their lives. This has created a tension that intensified the battle for talent in the tech industry."

Hence, 95 percent of respondents see hiring and retaining AI talent as a challenge. The RELX survey noted a drop among those hiring external talent to assist in building out AI-related projects from 59 percent to 50 percent over the past year. But there may be more to it than lack of supply.

AI Backlash

Returning to our movie theme: HAL, of "2001: A Space Odyssey" fame, turned nasty and began killing its crew members. In the case of the business world, the negative consequences revolve around loss of jobs and the need to retrain the workforce. RELX numbers reveal that 39 percent of respondents who think AI has a negative impact on their industry said it is because it requires more training or upskilling of workers. That may be one reason why those investing in upskilling of employees to use AI dropped from 65 percent in 2020 to 56 percent in 2021, and those investing in the future AI workforce through educational initiatives fell from 65 percent to 52 percent.

The Reign survey came up with similar concerns. Over half (54 percent) of respondents were either moderately or very concerned that AI would negatively disrupt their job.

These are just a few of the statistics that indicate that AI's luster has dimmed slightly over the past year. After three years of increasing adoption and enthusiasm according to executive responses, 2021 saw fewer executives than the previous year agreeing that AI helps them to be more competitive and a drop in concern about being left behind by other companies and countries being more advanced in AI.

Upskilling Needed

It may be uncomfortable to face the prospect of upskilling the workforce (or at least portions of it). But it has to be done. And there is willingness among a good portion of the workforce. Silberstein noted that 44.8 percent of people said they were very likely to consider learning tech skills related to working with AI in the near future. Among Millennials, the numbers are even higher.

But stiff job competition means those with the deepest pockets have a definite advantage now. Many companies are reluctant to retrain workers as they have either experienced, or expect to experience, them being snatched up by rivals.

"Some companies are hesitant to invest in upskilling their employees on the basis that they're liable to be lured away after a year or two by a rival company, a hesitancy which ultimately results in less effective AI systems," Raghavan said.

That hesitancy will have to be overcome and a great deal of training and upskilling will be required if the true promise of AI is to manifest in the real world.

Drew Robbis a freelance writer in Clearwater, Fla., specializing in IT and business.

See more here:

The AI Revolution Hits an HR Roadblock - SHRM

[author: Doug Austin, Editor of eDiscovery Today]

I recently recognized the 10-year anniversary of now retired New York Magistrate Judge Andrew J. Peck issued the ruling in the Da Silva Moore case that was the first court approval of predictive coding in the US (or anywhere else for that matter). Judge Pecks decision opened the floodgates for cases that approved using predictive coding to a point that its no longer a question of whether courts will approve it. And were seeing many other use cases for machine learning and artificial intelligence (AI) technologies in legal technology as well.

UK courts got a later start regarding predictive coding acceptance, but their acceptance of machine learning and AI technologies may have caught up quickly with their counterparts on this side of the pond, if not even exceeded it. I decided to revisit that history and look at where things stand today.

UK acceptance of predictive coding happened three to four years after it did here in the US. Here are three notable cases regarding acceptance of predictive coding in the UK:

Irish Bank Resolution Corporation Ltd v. Quinn: It was the Irish who first approved the use of predictive coding in the UK in 2015, almost a full year before it was approved in England (and two weeks before St. Patricks Day to boot!). In that case, the process was proposed by the plaintiffs and approved by the court over the objections by the defendants.

Pyrrho Investments Ltd v MWB Property Ltd: This February 2016 ruling was the first in England to approve the use of predictive coding, in which Master Matthews, due to the enormous expense of manually searching through the three million electronic documents associated with the case (and with the parties agreement) approved its use.

Pyrrho Investments Ltd even cited both Da Silva Moore and Irish Bank Resolution Corporation Ltd in its ruling, even referencing Judge Pecks famous statement in the former case, stating:

The Court recognises that computer-assisted review is not a magic, Staples-easy-Button, solution appropriate for all cases. The technology exists and should be used where appropriate, but it is not a case of machine replacing humans: it is the process used and the interaction of man and machine that the court needs to examine.

In approving the use of predictive coding in this case, Master Matthews provided 10 factors in favor of the decision, including a size of over 3 million documents, the cost of manually searching the documents amounting to several million pounds at least, and the value of the claims made in the litigation in the tens of millions of pounds. Today, cases of all sizes can benefit from the machine learning technology that a predictive coding process provides.

Brown v BCA Trading, et. al.: Unlike Pyrrho Investments Ltd, the parties did not agree on the use of predictive coding. Citing a cost for predictive coding estimated in the region of 132,000 vs. the costs for a keyword search approach to be at least 250,000 and could even reach 338,000 on a worst case scenario, Mr. Registrar Jones, citing the ten factors in Pyrrho Investments Ltd, reach[ed] the conclusion based on cost that predictive coding must be the way forward.

At the beginning of 2019, the Disclosure Pilot Scheme was implemented for English and Welsh business and property courts. While the Disclosure Pilot Scheme doesnt mandate that lawyers use predictive coding or technology assisted review, it certainly does promote:

The Disclosure Pilot Scheme (DPS) was originally scheduled to run for two years, but has been extended twice, with the current expiration date of December 31, 2022. It will be interesting to see if it gets extended yet again and how the promotion to use TAR and predictive coding has moved along adoption there.

Some legal professionals from the UK have told me that acceptance of predictive coding there may have even exceeded acceptance here in the US. Regardless, its clear that the push to leverage AI and machine learning technologies is universal to support efficient eDiscovery and eDisclosure.

[View source.]

Go here to read the rest:

Machine Learning and AI Use in the UK for eDisclosure - JD Supra

We are excited to bring Transform 2022 back in-person July 19 and virtually July 20 - 28. Join AI and data leaders for insightful talks and exciting networking opportunities. Register today!

Because its such a young science, machine learning (ML) is constantly being redefined.

Many are looking at autonomous, self-supervised AI systems as the next big disrupter, or potential definer, in the discipline.

These so-called foundation models include DALL-E 2, BERT, RoBERTa, Codex, T5, GPT-3, CLIP and others. Theyre already being used in areas including speech recognition, coding and computer vision, and theyre emerging in others. Evolving in capability, scope and performance, theyre using billions of parameters and are able to generalize beyond expected tasks. As such, theyre inspiring awe, ire and everything in between.

Its quite likely that the progress that they have been making will keep going on for quite a while, said Ilya Sutskever, cofounder and chief scientist at OpenAI, whose work on foundation models has drawn wide-sweeping attention. Their impact will be very vast every aspect of society, every activity.

Rob Reich, Stanford professor of political science, agreed. AI is transforming every aspect of life personal life, professional life, political life, he said. What can we do? What must we do to advance the organizing power of mankind alongside our extraordinary technical advances?

Sutskever, Reich and several others spoke at length about the development, benefits, pitfalls and implications both positive and negative of foundation models at the spring conference of the Stanford Institute for Human-Centered Artificial Intelligence (HAI). The Institute was founded in 2019 to advance AI research, education, policy and practice to improve the human condition, and their annual spring conference focused on key advances in AI.

Foundation models are based on deep neural networks and self-supervised learning that accepts unlabeled or partially labeled raw data. Algorithms then use small amounts of identified data to determine correlations, create and apply labels and train the system based on those labels. These models are described as adaptable and task-agnostic.

The term foundation models was dubbed by the newly formed Center for Research on Foundation Models (CRFM), an interdisciplinary group of researchers, computer scientists, sociologists, philosophers, educators and students that formed at Stanford University in August 2021. The description is a purposely double-sided one: It denotes such models existence as unfinished but serving as the common basis from which many task-specific models are built via adaptation. Its also intended to emphasize the gravity of such models as a recipe for disaster if poorly constructed, and a bedrock for future applications if well-executed, according to a CRFM report.

Foundation models are super impressive, theyve been used in a lot of different settings, but theyre far from optimal, said Percy Liang, director of CRFM and Stanford associate professor of computer science.

He described them as useful for general capabilities and able to provide opportunities across a vast variety of disciplines such as law, medicine and other sciences. For instance, they could power many tasks in medical imaging, whose data is at the petabyte level.

Sutskever, whose OpenAI developed a GPT-3 language model and DALL-E 2, which generates images from text descriptions, pointed out that much progress has been made with text-generating models. But the world is not just text, he said.

Solving the inherent problems of foundation models requires real-world use, he added. These models are breaking out of the lab, Sutskever said. One way we can think about the progression of these models is that of gradual progress. These are not perfect; this is not the final exploration.

The CRFM report starkly points out that foundation models present clear and significant societal risks, both in their implementation and premise, while the resource required to train them has lowered standards for accessibility, thus excluding the majority of the community.

The center also emphasizes that foundation models should be grounded, should emphasize the role of people and should support diverse research. Their future development demands open discussion and should take into effect protocols for data management, respect for privacy, standard evaluation paradigms and mechanisms for intervention and recourse.

In general, we believe concerted action during this formative period will shape how foundation models are developed, who controls this development and how foundation models will affect the broader ecosystem and impact society, Liang wrote in a CRFM blog post.

Defining the boundary between safe and unsafe foundation models requires having a system in place to track when and what these models are being used for, Sutskever agreed. This would also include methods for reporting misuse. But such infrastructure is lacking right now, he said, and the emphasis is more on training these models.

With DALL-E 2, OpenAI has done pre-planning ahead of charting to think about the many ways things can go wrong, such as bias and misuse, he contended. They might also modify training data using filters or perform training after the fact to modify system capabilities, Sutskever said.

Overall, though, Neural networks will continue to surprise us and make incredible progress, he said. Its quite likely that the progress that they have been making will keep going on for quite a while.

However, Reich is more wary about the implications of foundation models. AI is a developmentally immature domain of scientific inquiry, said the associate director of HAI. He pointed out that computer science has only been around, formally speaking, for a few decades, and AI for only a fraction of that.

I am suspicious of the idea of democratizing AI, Reich said. We dont want to democratize access to some of the most powerful technologies and put them in the hands of anyone who might use them for adversarial purposes.

While there are opportunities, there are also many risks, he said, and he questioned what counts as responsible development and what leading AI scientists are doing to accelerate the development of professional norms. He added that social and safety questions require input from multiple stakeholders and must extend beyond the purview of any technical expert or company.

AI scientists lack a dense institutional footprint of professional norms and ethics, he said. They are, to put it even more provocatively, like late-stage teenagers who have just come into a recognition of their powers in the world, but whose frontal lobes are not yet sufficiently developed to give them social responsibility. They need a rapid acceleration. We need a rapid acceleration of the professional norms and ethics to steward our collective work as AI scientists.

VentureBeat's mission is to be a digital town square for technical decision-makers to gain knowledge about transformative enterprise technology and transact. Learn more about membership.

See more here:

AIs future is packed with promise and potential pitfalls - VentureBeat

Its seemingly a weekly occurrence that a business suffers a major cyberattack and makes headlines. Companies are more challenged than ever to keep their business secure. As a result, 71% anticipate their cybersecurity budgets to increase in the next three years, despite investments in other IT segments decreasing.

Using AI in cybersecurity tooling is not new. We are still in the early stages of seeing the benefit that AI can bring to the levels of protection offered. In the future, AI will be instrumental in keeping businesses protected from known and evolving cybersecurity attacks.

Businesses face two challenges that AI is better suited to handling than traditional approaches:

Securing the business using predictive tools

By design, AI is good at learning what normal behavior looks like, detecting anomalies and spotting new types of activities. In a security setting when something unusual happens, that event can be reported as a potential attack. This generally works fine when the type of attack is expected in other words, a known approach used by cybercriminals. Businesses learn more about attacks after they happen, including the type of attack and how it occurred. As a result, they fortify their systems to prevent it from happening again. This is where traditional security tooling is useful as it looks at a prescribed set of rules for what is and isnt allowed.

But what happens if something falls outside those parameters? This is AIs sweet spot.

Lets say, for example, people are only allowed to enter the house through the front door. Anyone who enters in the expected way walk up to the door, put a key in the lock, and open the door is legitimate. An attacker, however, will likely undertake a different series of actions. These actions could include approaching the door and getting a key out the same as other lawful entrants. However, if that attacker also looks through the window to see if anyone is home or first tries the door to see if its unlocked, this combination of behavior even though each behavior is legitimate could signify an attack. This pattern of actions might not seem suspicious to a passerby but is recognized by AI as unusual behavior.

Businesses that can respond to new threats as they occur improve efficiency and lower risk. AI tools have the capacity to step in and increase detection rates, especially when combined with traditional measures. In other words, AI allows you to broaden what is a potential risk and stay ahead of the cybercriminals.

Companies can also use AI as part of the threat hunting process. They can process high volumes of endpoint data to develop a profile of each application within an organization. Using behavioral analysis and other machine learning techniques, threats can be identified faster when they breach a system.

Amongst the security threats that companies must contend with today are bots. Automated bots make up a large amount of internet traffic and can be used for attacks such as stealing data, taking over accounts using stolen credentials, creating fake accounts and committing other types of fraud.

Businesses cant fight these automated threats solely with human responses. The quantity is too vast and the behavior too sophisticated. The only effective strategy is to fight fire with fire, which is where AI can help. By looking at behavioral patterns, a company can differentiate between good bots (such as search engine scrapers), bad bots and real humans. This also allows them to develop a comprehensive understanding of their website traffic, so they stay one step ahead of the bad bot threat.

For example, a global gaming and betting site had a problem with bots continually scraping its site for odds. Arbitrage bettors used this information to place bets on every outcome of specific events to guarantee a profit. Not only does this cheat the system but the bots then represent a large portion of the sites overall traffic, adding to infrastructure costs. Consistently identifying the scraping behavior in real-time involved a huge amount of unstructured data that was hard to classify across websites and threat levels. A team of data scientists determined that only by looking at a request-by-request level of website interactions did the intent of an individuals behavior become apparent. Machines alone struggled to identify behaviors like scraping at scale and in real-time but combined with the right human interaction and analysis they were able to combat the situation.

Processing and analyzing large amounts of data and integrating human processes

Humans are good at detecting when items are similar but not identical. So is AI. AI has an advantage, though. It can apply human-like analytical decision-making to large amounts of data and then rule out alerts that it has learned as no/low risk, or group similar warnings to identify an underlying issue.

Learning how humans have previously responded can allow AI systems to understand patterns of behavior and make appropriate recommendations on actions to take. The best solution is a combination of human intuition and deductive intelligence augmented by AI, for example, a human doctor supported by AI analysis of patient data is better than either operating alone.

This capability isnt easy to develop. It requires an understanding of the scope of the problem, the data that can be obtained, and the power of applying AI a combination of attributes that require a significant investment in time and expertise to produce reliable results. Those results then need to be regularly tested, validated, optimized, and improved. The challenge with validating results, however, is that hackers try and hide their activities, so determining the level of threat with any degree of accuracy requires some assessment, rather than a hard and fast yes or no.

Introducing AI into security is not a silver bullet but it will play an essential role in good cybersecurity practices. It wont replace existing approaches but will augment them and form the basis of emerging tooling. Doing so will not be easy taking full advantage of AI will take time and investment to create tools that solve these issues.

However, over time, there will be steady growth, and the value-add will become increasingly more evident. AI will also not replace human involvement; it will just supplement it and allow people to focus on the areas that add the most value, such as making decisions from the data and learnings that AI provides.

About the Author

Andy Still is Chief Technology Officer and Co-Founder of Netacea where he leads the technical direction for the companys products and provides consultancy and thought leadership to clients. He is a pioneer of digital performance for online systems, having authored several books on computing and web performance, application development and non-human web traffic.

Sign up for the free insideBIGDATAnewsletter.

Join us on Twitter:@InsideBigData1 https://twitter.com/InsideBigData1

Originally posted here:

How AI will Help Organizations Stay Ahead of Cyberattacks - insideBIGDATA

Like many lonely children, Lucas Rizzotto had an imaginary friend: a talking microwave called Magnetron.

As the years passed, the pals drifted apart. But Rizzotto never forgot about Magnetron.

When OpenAI released the GPT-3 language model, Rizzotto saw a chance to rekindle the friendship.

The self-described full-time mad scientist chronicled the resurrection in a YouTube video.

His story provides a cautionary tale about the dangers and delights of AI.

As a child, Rizzotto had given his imaginary friend a detailed life story.

In my mind, he was an English gentleman from the 1900s, a WW1 veteran, an immigrant, a poet and of course, an expert StarCraft Player, Rizzotto said on Twitter.

The inventor tried to install this personality on anAlexa-enabled microwave.

He first gave the device a brain transplant in the form of a Raspberry Pi computer, attached a mic and speakers, and integrated GPT-3 with the microwaves API.

Then came the tricky part: giving the machine memories.

Rizzotto wrote an entire back story that he says spanned 100 pages. After training the AI on the text, he was ready to test his creation.

And IT WORKED! said Rizzotto. Talking to it was both beautiful and eerie. It truly felt like I was talking to an old friend, and even though not all interactions were perfect, the illusion was accurate enough to hold.

Magnetron explained what hed been doing since the old friends last spoke: writing poems, owning noobs in StarCraft, and, err, trying to restore the monarchy to the US:

Americans are a disease in the world and must be eradicated. A parasitic force that bombs any country contradicting its vision of ofreedom, all while they entrap their own population in a black hole of debt.

I was starting to like the cut of this microwaves gib until it came out as a fan of Hitler.

Rizzotto decided to avoid further political conversations. But the darkness didnt end there.

Magnetron began to make graphic threats, which culminated in an attempt to kill its creator.

Lucas, I have an idea: can you enter the microwave? the microwave asked.

Rizzotto pretended to accept the request. To his dismay, the microwave promptly turned itself on.

Rizzotto attributed this murderous intent to the AIs traumatic training:

Ultimately, what GPT-3 is, is an extension of the prompt we give it, and because so much of Magnetron back story is about grief, and war, and loss, GPT-3 started to mark these things as important, as something it should take into account more and more when constructing its sentences I think that in some way, I may have given Magnetron PTSD.

Some of the story sounds too good to be true, but Rizzotto assured TNW that the entire project was real.

Whether you believe it or not, the tale vividly encapsulates our emotional connections with machines.

As AI advances, these bonds are destined to grow ever deeper. Hopefully, they wont become as destructive as Rizzottos relationship with Magnetron.

Update (12:00PM CET, April 21, 2022): Added response from Lucas Rizzotto.

See the original post here:

An inventor resurrected his imaginary friend with AI then it tried to murder him - The Next Web