Mediaboss Marketing

Last week an artificial intelligence called NooK beat eight world champion players at bridge. That algorithms can outwit humans might not seem newsworthy. IBMs Deep Blue beat world chess champion Garry Kasparov in 1997. In 2016, Googles AlphaGo defeated a Go grandmaster. A year later the AI Libratus saw off four poker stars. Yet the real-world applications of such technologies have been limited. Stephen Muggleton, a computer scientist, suggests this is because they are black boxes that can learn better than people but cannot express, and communicate, that learning.

NooK, from French startup NukkAI, is different. It won by formulating rules, not just brute-force calculation. Bridge is not the same as chess or Go, which are two-player games based on an entirely known set of facts. Bridge is a game for four players split into two teams, involving collaboration and competition with incomplete information. Each player sees only their cards and needs to gather information about the other players hands. Unlike poker, which also involves hidden information and bluffing, in bridge a player must disclose to their opponents the information they are passing to their partner.

This feature of bridge meant NooK could explain how its playing decisions were made, and why it represents a leap forward for AI. When confronted with a new game, humans tend to learn the rules and then learn to improve by, for example, reading books. By contrast, black box AIs train themselves by deep learning: playing a game billions of times until the algorithm has worked out how to win. It is a mystery how this software comes to its conclusions or how it will fail.

NooK nods to the work of British AI pioneer Donald Michie, who reasoned that AIs highest state would be to develop new insights and teach these to humans, whose performance would be consequently increased to a level beyond that of a human studying by themselves. Michie considered weak machine learning to be just improving AI performance by increasing the amount of data ingested.

His insight has been vindicated as deep learnings limits have been exposed. Self-driving cars remain a distant dream. Radiologists were not replaced by AI last year, as had been predicted. Humans, unlike computers, often make short work of complicated, high-stake tasks. Thankfully, human society is not under constant diagnostic surveillance. But this often means not enough data for AI is available, and frequently it contains hidden, socially unacceptable biases. The environmental impact is also a growing concern, with computing projected to account for 20% of global electricity demand by 2030.

Technologies build trust if they are understandable. Theres always a danger that black box AI solves a problem in the wrong way. And the more powerful a deep-learning system becomes, the more opaque it can become. The House of Lords justice committee this week said such technologies have serious implications for human rights and warned against convictions and imprisonment on the basis of AI that could not be understood or challenged. NooK will be a world-changing technology if it lives up to the promise of solving complex problems and explaining how it does so.

Read the rest here:
The Guardian view on bridging human and machine learning: its all in the game - The Guardian

Rapidly developing artificial intelligence (AI) technology is becoming increasingly critical for innovation and economic growth. To secure American leadership and competitiveness in this emerging field, policymakers should create an innovation-friendly environment for AI research. To do so, federal authorities should identify ways to engage the private sector and research institutions.

The National AI Research and Development (R&D) Strategic Plan, which will soon be updated by the Office of Science and Technology Policy (OSTP) and the National Science and Technology Council (NSTC), presents such an opportunity. However, the AI Strategic Plan needs several updates to allow the private sector and academic institutions to become more involved in developing AI technologies.

First, the OSTP should propose the creation of a federal AI regulatory sandbox to allow companies and research institutions to test innovative AI systems for a limited time. An AI sandbox would not only benefit consumers and participating companies; it would also enable regulators to gain first-hand insights into emerging AI systems and help craft market-friendly regulatory frameworks and technical standards. Regulators could also create sandbox programs to target innovation on specific issuessuch as human-machine interaction and probabilistic reasoningthat the AI Strategic Plan identifies as priority areas in need of further research.

Second, the updated AI strategy should outline concrete steps to publish high-quality data sets using the vast amount of non-sensitive and non-personally identifiable data that the federal government possesses. AI developers need high-quality data sets on which AI systems can be trained, but the lack of access to these data sets remains a significant challenge for developing novel AI technologies, especially for startups and businesses without the resources of big tech companies. The costs associated with creating, cleaning, and preparing such data sets are too high for many businesses and academic institutions. For example, AlphaGo, a software produced by Google subsidiary DeepMind, made headlines in March 2016 when it defeated the human champion of a Chinese strategy game. More than $25 million was spent on hardware alone to train data sets for this program.

Recognizing this challenge, the AI Strategic Plan recommended the development of shared public data sets, but progress in this area appears to be slow. Under the 1974 Privacy Act, the U.S. government has not created a central data repository, which is important due to the privacy and cybersecurity risks that such a repository of sensitive information would pose. However, different U.S. agencies have created a wide range of non-personally identifiable and non-sensitive data sets intended for public use. Two notable examples are the National Oceanic and Atmospheric Administrations climate data and NASAs non-confidential space-related data. Making such data readily available to the public can promote AI innovation in weather forecasting, transportation, astronomy, and other underexplored subjects.

Therefore, the AI strategy should propose a framework that enables the OSTP and the NSTC to work with government agencies in order to ensure that non-sensitive and non-personally identifiable dataintended for public useare made available in a format suitable for AI research by the private sector and research institutions. To that end, the OSTP and the NSTC could use the federal governments existing FedRAMP classification of different data types to decide which data should be included in such data sets.

Finally, the AI Strategic Plan would benefit from a closer examination of other countries AI R&D strategies. While policymakers should exercise caution in making international comparisons, awareness of these broader trends can help the United States capitalize on different countries successes and avoid their regulatory mistakes. For example, the British and French governments recently spearheaded initiatives to promote high-level interdisciplinary AI research in multiple disciplines. Likewise, the Chinese government has launched similar initiatives to encourage cross-disciplinary academic research at the intersection of artificial intelligence, economics, psychology, and other disciplines. Studying and evaluating other countries approaches could provide American policymakers insights into which existing R&D resources should be devoted to interdisciplinary AI projects.

To maximize the benefit of this comparative approach, the AI Strategic Plan should propose mechanisms to conduct annual reviews of the global AI research and regulatory landscape andevaluations of its successes and failures.

Ultimately, due to AIs general-purpose nature and its diffusion across the economy, the AI Strategic Plan should focus on enabling a wide range of actors, from startups to academic and financial institutions, to play a role in strengthening American AI innovation. An innovation-friendly research environment and an adaptable, light-touch regulatory approach are vital to secure Americas global economic competitiveness and technological innovation in artificial intelligence.

Ryan Nabil is a Research Fellow at the Competitive Enterprise Institute in Washington, DC.

Image: Flickr/U.S. Air Force.

Original post:
How to Strengthen America's Artificial Intelligence Innovation - The National Interest

The Future of Life Institute (FLI) was founded in March 2014 by eminent futurologists and researchers to reduce catastrophic and existential risks to humankind from advanced technologies like artificial intelligence (AI). Elon Musk, who is on FLI's advisory board, donated $10 million to jump-start research on AI safety because, in his words, 'with artificial intelligence, we are summoning the devil'. For something that everyone is singing hosannas to these days, and treating as a solution to almost all challenges faced by industry or healthcare or education, why this cautionary tale?

AI's perceived risk isn't only from autonomous weapon systems that countries like the US, China, Israel and Turkey produce that can track and target humans and assets without human intervention. It's equally about the deployment of AI and such technologies for mass surveillance, adverse health interventions, contentious arrests and the infringement of fundamental rights. Not to mention about the vulnerabilities that dominant governments and businesses can insidiously create.

AI came into global focus in 1997 when IBM's Deep Blue beat world chess champion Garry Kasparov. We came to accept that the outcome was inevitable, considering it was a game based on logic. And that the ability of the computer to reference past games, figure options and select the most effective move instantly, is superior to what humans could ever do. When Google DeepMind's AlphaGo program bested the world's best Go player Lee Sedol in 2016, we learnt that AI could easily master games based on intuition too.

AI, AI, SirAs the United Nations Educational, Scientific and Cultural Organisation (Unesco) sharpened the focus in recognising the ethical dilemmas that AI could create, it has embarked on developing a legal, global document on the subject. Situations discussed include how a search engine can become an echo chamber upholding real-life biases and prejudices - like when we search for the 'greatest leaders of all time', and get a list of only male personalities. Or the quandary when a car brakes to avoid a jaywalker and shifts the risk from the pedestrian to the travellers in the car. Or when AI is exploited to study 346 Rembrandt paintings pixel by pixel, leveraging deep-learning algorithms to produce a magnificent, 3D-printed masterpiece that could deceive the best art experts and connoisseurs.

Then there is the AI-aided application of justice in legislation, administration, adjudication and arbitration. Unesco's quest to provide an ethical framework to ensure emerging technologies benefit humanity at large is, indeed, a noble one.

Interestingly, computer scientists at the Vienna University of Technology (TU Wein), Austria, are studying Indian Vedic texts, and applying them to mathematical logic. The idea is to develop reasoning tools to address deontic - relating to duty and obligation - concepts like prohibitions and commitments, to implement ethics in AI.

Logicians at the Institute of Logic and Computation at TU Wein and the Austrian Academy of Science are also gleaning the Mimamsa, which interprets the Vedas and suggests how to maintain harmony in the world, to resolve many innate contradictions. Essentially, as classical logic is less useful when dealing with ethics, deontic logic needs to be developed that can be expressed in mathematical formulae, creating a framework that computers can comprehend and respond to.

Isaac Asimov's iconic 1950 book, I, Robot, sets out the three rules all robots must be programmed with: the Three Laws of Robotics - 1. To never harm a human or allow a human to come to harm. 2. To obey humans unless this violates the first law. 3. To protect its own existence unless this violates the first or second laws. In the 2004 film adaptation, a larger threat is envisaged - when AI-enabled robots rebel and try to enslave and control all humans, to protect humanity for its own good, by their dialectic.

Artificially RealIn the real world, there is little doubt that AI has to be mobilised for the greater good, guided by the right human intention, so that it can be leveraged to control larger forces of nature like climate change and natural disasters that we can't otherwise manage. AI must be a means to nourish humanity in multifarious ways, rather than unobtrusively aid its destruction. It is obvious that the Three Laws of Robotics must be augmented, so that expanded algorithms help the AI engine respect privacy, and not discriminate in terms of race, gender, age, colour, wealth, religion, power or politics.

We're seeing the mainstreaming of AI in an age of exponential digital transformation. How we figure its future will shape the next stage of human evolution. The time is opportune for governments to confabulate - to shape equitable outcomes, a risk management strategy and pre-emptive contingency plans.

More here:
Why it's time to address the ethical dilemmas of artificial intelligence - Economic Times

The DeepMind Academic Fellowship Program provides an opportunity for early-career researchers in the fields of Computer Science and Artificial Intelligence to pursue postdoctoral study and build the experiences and research profile that will enable them to progress to full academic or other research leadership roles in future.

Alongside financial support, DeepMind provides opportunities for fellows to be mentored by senior DeepMind researchers. DeepMind will not direct their research and fellows are free to pursue any research direction they wish.

Fellowships are open to early-career researchers who have completed a PhD in Machine Learning, Computer Science, Statistics or another relevant field by the time they start their postdoc. We particularly encourage candidates who identify as Black to apply because this group is currently underrepresented in AI research.

DeepMind has partnered with University of Cambridge, University College London and Queen Mary University of London to launch the Fellowship program in 2021. Application for the 2nd cohort of the program is expected to open later in 2022.

Read more from the original source:
About - Deepmind

In their 2009 manifesto, Neural-Symbolic Cognitive Reasoning, Artur Garcez and Luis Lamb, discussed the 1990s popular idea of integrating neural networks and symbolic knowledge. They cited Towell and Shavliks KBANN, Knowledge-Based Artificial Neural Network that uses a system to insert rules, refine and extract data from neutral network; a model empirically proving to be effective. Industry leaders, including contingencies at IBM, Intel, Google, Facebook, and Microsoft, and researchers like Josh Tenenbaum, Anima Anandkumar, and Yejin Choi, are starting to apply this technique in 2022. The recent AI developments, challenges, and stagnancy is leading the industry to consider this hybrid approach to AI modelling.

Neuro-Symbolic AI is essentially a hybrid AI leveraging deep learning neural network architectures and combining them with symbolic reasoning techniques. For example, we have been using neural networks to identify the shape or colour a particular object has. Applying symbolic reasoning to it can take it a step further to tell more exciting properties about the object, such as the area of the object, volume and so on.

AI has been the talk of the town for more than a decade, and while it has stood true to several promises, the majority of the claims are still to be met, and the challenges connected to AI have only been increasing. In the past year, GPT-3 has asked individuals to commit suicide, Alexa has challenged a ten-year-old to touch a coin to the plug, and Facebooks algorithm has identified a man of colour as a primate. This is not any different from Microsofts Tay asserting Hitler was right or Ubers self-driving cars crossing red lights a few years ago. With every GPT-3 development, we have seen a downfall. The present efforts to ensure explainable, fair, ethical and efficient AI need to be supported by changes in how we approach artificial intelligence.

Scientist and AI author and entrepreneur Gary Marcus recently wrote about deep learning hitting a wall and the responsibility of AGI. It must be like stainless steel, stronger and more reliable and, for that matter, easier to work with than any of its constituent parts. No single AI approach will ever be enough on its own; we must master the art of putting diverse approaches together if we are to have any hope at all.

Since AI use learning and reasoning in a quest to be like humans, the neuro-symbolic approach allows us to combine these strengths to make inferences based on the existing neural networks and learn through symbolic representations. Knowledgeable Magazine asserted this hybrid to show duckling-like abilities. Ducklings can imprint colours and shapes and differentiate between them. Moreover, they can differentiate between same and different, an aspect AI still struggles with. Here, symbolic AI would involve symbols for physical objects and colours in its knowledge base. The base also consists of general rules to differentiate between them. This, combined with the deep nets, allows the model to be more efficient.

The combinations demand humans supply a knowledge base/symbolic rules for the AI to leverage while the automated deep nets find the correct answers. The hybrid uses deep nets, instead of humans, to generate only those portions of the knowledge base that it needs to answer a given question, notes Knowledgeable Magazine.

DeepMind has seen some of the best success with their board-game playing AI models Go, Chess, MuZero and more. These are hybrid models using symbolic AI. For instance, AlphaGo used symbolic-tree search with deep learning, AlphaFold2 combines symbolic ways of representing the 3-D physical structure of molecules with the data-trawling characteristics of deep learning. Deepmind has asserted the qualities of symbolic learning in AI in a recent blog post. This approach will allow for AI to interpret something as symbolic on its own rather than simply manipulate things that are only symbols to human onlookers, they said. This allows for AI with human-like fluency. IBM has asserted neuro- symbolic AI is getting AI to reason. The LNN technique, Logical Neural Network, was introduced and created on the foundations of deep learning and symbolic AI. Given the combination, the software can successfully answer complex questions with minimal domain-specific training.

Two major conferences have been held asserting this needthe 2019 Montreal AI Debate between Yoshua Bengio and Gary Marcus and the AAAI-2020 fireside conversation with Laureate Daniel Kahneman, Geoffrey Hinton, Yoshua Bengio and Yann LeCun. The key takeaway from these events was the need for AI to have a reasoning layer with deep learning to frame a rich future of AI.

Excerpt from:
Experts believe a neuro-symbolic approach to be the next big thing in AI. Does it live up to the claims? - Analytics India Magazine