Yoshua Bengio on Bridging Biological and Artificial Intelligence: Challenges, Learning Paradigms, and Ethical AI

The discussion centers on the fundamental differences and potential synergies between biological and artificial neural networks, particularly highlighting the mystery of how biological systems achieve credit assignment over very long time spans. Bengio argues that current deep learning models, despite their successes, possess a "low-level" and "not robust" understanding of the world, lacking the abstract and causal reasoning capabilities of humans. He emphasizes the need for new training objectives and frameworks that move beyond passive observation to active learning through intervention, akin to how children learn by interacting with their environment. A crucial distinction is drawn between merely increasing network depth or size and fundamentally changing learning paradigms. Bengio believes that simply scaling up current architectures won't solve the core issues of shallow understanding and poor generalization. He advocates for "disentangled representations" where important causal factors are separated, and, more profoundly, "disentangled mechanisms" (rules) that relate these variables, preventing catastrophic forgetting and enabling better generalization to out-of-distribution data. This contrasts with classical AI's symbolic, rule-based systems which struggled with unconscious knowledge and uncertainty, but offered a form of compositionality that current neural nets lack. For immediate progress, Bengio suggests focusing on training objectives that reward exploration and causal understanding, and jointly learning from language and world models. He also highlights the opportunity for academic research to tackle "simple" environments where current state-of-the-art methods still fail, requiring millions of examples where humans need dozens. Regarding AI bias, he recommends short-term techniques like adversarial methods to reduce discrimination in datasets, advocating for government regulation to enforce their use even if it slightly reduces prediction accuracy. Long-term, he is interested in instilling moral values into computers, starting with detecting basic emotions like anger in response to unfair situations. The conversation touches on the broader societal implications of AI, distinguishing between short-to-medium term concerns (security, job market, power concentration, discrimination, autonomous weapons) and the "very unlikely" existential risk scenarios often depicted in science fiction like *Ex Machina*. Bengio criticizes such portrayals for misrepresenting the collaborative, incremental nature of scientific progress. He stresses the importance of instilling moral values and teaching strategies into AI, viewing "machine teaching" as a critical, underexplored area for future human-machine interaction, and emphasizes the need for diverse research directions and open debate within the AI community, rather than a singular approach.

one of the things I studied recently something that we don't know how biological neural networks do but would be really useful for artificial ones is the ability to do credit assignment through very long time spans

current state-of-the-art neural nets trained on large quantities of images or texts have some level of understanding of you know what explains those datasets but it's very basic it's it's very low-level and it's not nearly as robust and abstract in general as our understanding

I think the crucial thing is more the training objectives the training frameworks for example going from passive observation of data to more active agents which learn by intervening in the world the relationships between causes and effects

I don't think that having more more depth in the network in the sense of instead of a hundred layers we have ten thousand is going to solve our problem

even our state-of-the-art deep learning methods fail to learn models that understand even very simple environments like some Gridworlds that we have built even these fairly simple environments I mean of course if you train them with enough examples eventually they get it but it's just like instead of what instead of what humans might need just dozens of examples these things will need millions right for very very very simple tasks

there is something really powerful that comes from distributed representations the thing that really makes neural Nets work so well and it's hard to replicate that kind of power in a symbolic world

I'm hypothesizing that in the right high-level representation space both the variables and how they relate to each other can be disentangled and that will provide a lot of generalization power

I think the the picture of terminator and you know AI lose and killing people and super intelligence that's gonna destroy us whatever we try isn't really so useful for the public discussion because for the public discussion that things I believe really matter are the short-term and mini term very likely negative impacts of AI on society

science is not happening in some hidden place by some you know really smart guy one person one person this is totally unrealistic this is not how it happens even a team of people in some isolated place will not make it science moved by small steps thanks to the collaboration and community of a large number of people interacting

I think it's time that machine learning people look at this because in the future we'll have more and more human machine interaction with a human in the loop and I think understanding how to make this work better all the problems around that are very interesting and not sufficiently addressed