Author: Peter Rudin

Research to Overcome today’s Limits of AI

Posted by Peter Rudin on 24. January 2020 in Essay No Comments

The obsession with creating bigger datasets and bigger neural networks has side-lined some of the important questions and areas of research regarding AI. We need new models to advance AI.

Artificial neural networks rely on the point model, treating neurons as nodes that tally inputs and pass the sum through an activity function. Neuroscientists have discovered that dendrite compartments which are part of a neuron can also perform computations that mathematicians had categorized as unsolvable.

Lack of causality is one of the shortcomings of current machine learning systems. Systems that compose and manipulate named objects and semantic variables with causal structures will overcome these limits of AI.

Read More

Embodied AI: Genetic vs. Experience Based Intelligence

Posted by Peter Rudin on 17. January 2020 in News No Comments

In a recent debate about the future of AI between Gary Marcus, Professor Emeritus in the Department of Psychology at NYU and Yoshua Bengio, Professor at the Department of Computer Science at the University of Montréal, the question came up of how much information the human genome can encode.

This is a hot topic in AI these days, as people debate how much prior knowledge needs to be pre-wired into AI systems, in order to get them to achieve something more akin to natural intelligence.

Making a true AI may require building something that has to do more than solve some specific computational problems.  It must be embodied.

Read More

The Evolution of Artificial Collective Intelligence (ACI)

Posted by Peter Rudin on 10. January 2020 in Essay No Comments

Members of the first human groups shared the instinct to combine their respective information and expertise to meet survival tasks they could not possibly meet individually. Those early forms of collective intelligence (CI) gave rise to language and tools which, in turn, enabled new forms of collective intelligence that could absorb more existential complexity.

According to Thomas Malone, Director of the Center for Collective Intelligence at MIT, web-based software tools allow people to interact and collaborate in new ways. The relationship between CI and AI defines ACI as a new research area with a few major drivers.

One issue that must be solved relates to the problem of digital identity. We must create an identity system in which human beings can control their identity.

Read More

Do You Want to Create an AI-Copy of Yourself?

Posted by Peter Rudin on 3. January 2020 in News No Comments

“We each have more potential than ever to solve humanity’s biggest challenges and leave meaningful legacies through the power of AI. But our AI must be controlled and owned by individuals – not a few corporations”.

“By making it possible for everyone to have their own AI, we’re putting control back in the hands of individuals and unlocking the limits of humanity”.

The potential impact of these statements, posted on the Website of the AI-Foundation, can be experienced by an App that is being released in 2020. Eventually, the Foundation plans to release a tool for anyone to record and build their own AI.

Read More

AI and the Challenges ahead, a critical Assessment

Posted by Peter Rudin on 27. December 2019 in Essay 1 Comment

While educational institutions are overwhelmed by an onslaught of new students reaching for a degree in machine-learning, there is growing concern among members of the AI community that machine-learning and deep neural networks (DNNs) are flawed with severe problems.

Most advances in the field are associated with creating bigger neural networks and training them with more and more data. Excitement has blinded research to one of the fundamental problems that AI technology still suffers from: CAUSALITY.

DNN algorithms are powerful, but to think that they ‘think and learn’ in the same way as humans do is incorrect.

Read More

How AI Can Transform Psychiatry

Posted by Peter Rudin on 20. December 2019 in News 1 Comment

The rapid embracing of artificial intelligence in psychiatry has a flavor of being the current “wild west”; a multidisciplinary approach that is very technical and complex yet seems to produce findings that resonate.

Explainability, transparency, and generalizability are critical for establishing the viability of using artificial intelligence in psychiatry. Defining these three issues helps towards building a framework to ensure trustworthiness.

“We found that the computer’s AI models can be at least as accurate as clinicians,” says Peter Foltz, a researcher from the University of Colorado at Boulder.

Read More

Can we Grade AI and Compare it with Human Intelligence?

Posted by Peter Rudin on 13. December 2019 in Essay No Comments

With the resurgence of AI in the late nineties, applying Deep Learning to solve specific cognitive problems, stipulates the question what constitutes intelligence? How can one create intelligence artificially if we have no precise definition of what intelligence is?

Applying a framework like Francois Chollet’s Abstraction and Reasoning Corpus (ARC) is likely to empower AI research with a new perspective on defining and evaluating intelligence.

Measuring AI performance against human intelligence will eventually advance AI to the point where artificial cognitive intelligence will become a commodity, providing intelligence as a service to everyone.

Read More

How Does Language Emerge?

Posted by Peter Rudin on 6. December 2019 in News No Comments

How did the almost 6,000 languages of the world come into being? Researchers have tried to simulate the process of developing a new communication system in an experiment — with surprising results.

People create reference to actions and objects via signs that resemble things. The prerequisite for this is a common ground of experience between interaction partners. Partners also coordinate by imitating each other such that they use the same signs for the same things.

The studies demonstrate that communication cannot be reduced to words alone. When there is no way to use conventional spoken language, people find other ways to get their message across.

Read More

From Data-Centric to Human-Centred AI

Posted by Peter Rudin on 29. November 2019 in Essay No Comments

In his most recent book, “Human Compatible”, Stuart Russell takes on the challenges of AI to issues such as human purpose, authority and basic wellbeing. The major challenge ahead is to move AI from a data-centric to a human-centred approach, also referred to the creation of a General-Purpose AI (GAI).

Leveraging data and people’s expertise in new ways offers a path forward for smarter decisions, more innovative policymaking, and more accountability in governance.

Collective Intelligence, augmented by AI, is likely to enhance the development of human wisdom, enriching our individual life vis-à-vis the ever-growing complexity of scientific discovery and its impact on our society.

Read More

A Catalogue of Skills AI Can Handle Today

Posted by Peter Rudin on 22. November 2019 in News No Comments

(AI) being an increasingly present part of our everyday lives. But, many of us would be quite surprised to learn of some of the AI-skills that are designed to provide convenience or improve human functionality:

Read and write; hear, understand and speak;
see, touch and move; play games, debate and create;
understand emotions and read your mind.

Many of these skills are steadily improved by ongoing AI-research. However, the human capacity to associate and to cross-relate skills remains unchallenged so far. Going from ‘Narrow AI’ to ‘Artificial General Intelligence’ will create new scenarios and challenges for our society at large.

Read More