AI That Understands the World, Using Probabilistic Programming | Vikash Mansinghka | TEDxMIT

This talk shows how to make smarter, safer AI that understands the world like we do, using a new symbolic medium that I helped to invent, called probabilistic programming. Thanks to recent developments at MIT, probabilistic programming has begun to scale in surprising ways, outperforming machine learning algorithms. It also has yielded the new explanation of how the spiking of biological neurons can implement the risk-sensitive guesses and bets that make up our perceptions and thoughts. This human-like AI technology has implications for how we understand the world and ourselves. "
Vikash Mansinghka is a Principal Research Scientist at MIT, where he leads the Probabilistic Computing Project, part of MIT's CSAIL, Department of Brain & Cognitive Sciences, and the Quest for Intelligence. Vikash holds S.B. degrees in Mathematics and in Computer Science from MIT, as well as an M.Eng. in Computer Science and a PhD in Computation. He also held graduate fellowships from the National Science Foundation and MIT’s Lincoln Laboratory. His PhD dissertation on natively probabilistic computation won the MIT George M. Sprowls dissertation award in computer science, and his research on the Picture probabilistic programming language won an award at CVPR. He co-founded three VC-backed startups: Prior Knowledge (acquired by Salesforce in 2012) and Empirical Systems (acquired by Tableau in 2018), and Common Sense Machines (funded in 2020), and advises in companies such as DeepMind, Intel, and Google. He served on DARPA’s Information Science and Technology advisory board from 2010-2012, currently serves on the editorial board for the Journal of Machine Learning Research, and co-founded the International Conference on Probabilistic Programming." This talk was given at a TEDx event using the TED conference format but independently organized by a local community. Learn more at www.ted.com/tedx

[@CoinEZ]

I love the concept so much

[@giovannimazzocco499]

Beautiful talk. Loved the example in Gen. PPL is the future.

[@kipropcollins4220]

Now that is how to get me hooked!!!

[@GaninduNanayakkara]

Nice!!!

[@MrMentalpuppy]

Cycorp?

[@francisdelacruz6439]

The age of GPT is the age of AI failing so bad it doesn’t know it failed. So on to more productive theories.

[@RyluRocky]

What? No, both machine learning and Probability algorithms are separate from GPT which is an LLM, the only reason the probability approach can do half of the amazing things it’s doing is because GPT is interpreting all of the symbolic meaning. Also the Tesla cars don’t use LLMs at all, it mostly uses neural networks in machine learning. Transformers (which is a big part of LLMs) is the biggest discovery in History, It is the human brain to the body as it is the the driver to Intelligence, everything before LLMs that we called AI (artificial intelligence) wasn’t AI, by far the closest at the time was neural networks (which is still an integral piece of LLMs) while everything else was complex pattern algorithms merely mimicking techniques whereas an LLM is an actual brain that can use those techniques it literally understands information in relation to other information. If you have something that can understand information everything else intelligence wise will fall into place.