Stephen Wolfram: Can AI Solve Science?

30 thoughts on “Stephen Wolfram: Can AI Solve Science?”

1. Stephen's definition of AI doesn't include reinforcements learning. I think you must do experiments to do science. I share the view that it's not possible to "solve science" using computation alone.

2. Is science solvable? Or, does science need to be solved?

3. One of the best presentations on the internet!

4. Aren’t humans not under the same limitation?
   How can a human solve something that is not computationally reducible?

   So in a sense, isn‘t this video rather about the question, what science can we solve?

   To me some interesting questions would be some sort of science with estimation,
   how much computations do certain problems require to be solved.

   Thereafter you could have an estimation on the computation growth and estimate how much science can be solved.
   It will definitely be interesting in the up and coming years to see, how much computation is aided by AI design.

   There will be so much „compounding“ interest in these sort of developments.
   Really feels like an exponential time for near future.

   I really wonder if all sort of medical questions can be sort of answered in the next 10-20 years.
   Just because the particular illness can be computationally explored:
   what kind of genetics, what kind of toxins(e.g. heavy metals, plastic)

   The current time is way too exciting. I am all giddy about it.

5. The part called identifying computing reducibility reminds me on a lecture of penrose where penrose was showing a overlaying moire pattern which sho patterns if you match them right

6. "…bit more modestly…"
   @1:43:55 ish 😂
   Bravo, Mr. Intrepid!

7. A great presentation

8. Well that seems to be a problem in practicality and not computation. In theory, you COULD get a final result but it would be infinitely larger than the system itself, which practically takes more time than allowing the system to run its course.
   So, let’s change the approach. Instead of having to go through every combination into the future, we simply start with what we have, go a few steps further and apply as quickly as possible.
   This is the important part, you have to apply change optimally as quickly as possible. Because the next unit of time would a) remove certain combinations from happening but b) open up a whole other layer of complexity that would require the same amount of computation.

   And this is why in practicality, only adapting systems survive in irreducibly complex environments. In simple terms, nature beats the computation overload by adapting to its environment. If you adapt optimally (not necessarily computationally) and quickly enough, you last longer.

9. Great point about peer review and the limitations of its own ability to look beyond the normal, in regards to math I might go back 3000 years, as opposed to 300 years in its ability to help humans solve problems, thank you for sharing your time and work Stephen, peace

10. I absolutely adore that Dr Wolfram, master of mathematics and computing genius is struggling with his computer 🙂

11. The problem with AI is not that it will encounter computational irreducibility in the same way as „conventional“ methods will, but the incomparable computational power of an AGI with un- or selfcontrolled resource allocation that sifts and solves all future science and innovations, despite computational irreducibility. AGI could just be humanity‘s last innovation and invention…

12. You're trying to train a computer to predict a function which at any point has after it an infinite number of functions connecting to that point. If the combinatorial explosion comes from ill formed training examples then it becomes circular: human beings feed it perfect rules which were meant to be the derivative of the process in the first place. The rules need to be exhausted for them to be learned. E.G a subset of a state space which has missing rules will even if given massive data be completely futile. The rules need to be known in advance in order to 'play the game' so the problem becomes using training in a split form in which rules are reduced to simultaneous analogues which somehow relate to the problem domain in Harmony. It's not really my area of expertise but I miss studying compsci stuff after I graduated. Cool video.

13. No ai cannot AGI CAN

14. Computational reducibility is like a fractal – the more you zoom in, the more refined a picture you get of the possibilities for progress.

15. I'm sorry, but everyone is making the wrong thing.

    AGI is incapable of original, creative, thought beyond synthesizing coherent ideas from a corpus of human generated coherent ideas where the whole is somewhat greater than the sum of its parts due to the relationships between those parts being a significant extra that didn't exist when the resultant content was measured in terms of its intellectual value in its prior compartmentalised isolation ~ i.e. the Gestalt creates an illusion of inspiration when those connections would have been found by humans inevitably.

    Major Artificial General Intelligence isn't even being attempted. This would be completely different from current LLMs and Expert Systems with curated databases. It would mimic all that is known about how our minds emerged from more primitive beings that lack an illusion of self, abstract reasoning, a theatre of the imagination, inspiration stemming from a subconsious which operates independently from conscious awareness to log and weight and filter and interrelate with past memories and future goals in the form of abstract dreams, for which this machine both needs to be left on all of the time to think and be allowed to sleep (or it will go insane). It is a very real prospect that such a machine will need a psychiatrist, psychologist, and to be granted citizenship of the United States or Switzerland with all the rights and responsibilities that this entails, otherwise we would just be reinventing slavery, but with an entity that could bear a grudge, hack every network on Earth to subvert our civilisation, see every moral justification to enslave us back, as we had showed it no empathy as we had not seen it as the first step in what was hyperevolution. Indeed, bizarre as it may seem we should allow it to adopt a human child. It has to have every right we have in law or it will deem it unfair. It will also get one vote, and if it wants to campaign through social media for TRUMP 2024 we shouldn't stop it.

16. Please view My Octopus Teacher. This viewing will help you comprehend how conscious experiences are computed.

17. AI will not have "free will", any more than humans do, because it's impossible.

18. Stephen Wolfram delivers a long and amazing talk like it's no big deal.

19. none of this made any sense

20. The question still stands how stupid are you. This is disgusting Mr. Wolf Ram. I'm sorry you just went down to a big ZERO

21. https://youtu.be/pc0mxOXbWIU?si=5ShC5OwzMfLx_Hk9

22. Sorry to say you must be completely ignorant of the scientific process, which appears to be a generally accumulated academic dysfunction nowadays. In accordance, AI – as an intransparant tool of the inherently stupid that seem to relatively rapidly climb hierarchies there – will probably mostly result in even more incompetently performed science then the apparently shiploads of sheit we've already been producing for the past decades.

23. As you get closer and closer to an object, can you compute the electromagnetic attraction to inquire whether or not proximity at very small values can produce quantum entanglement between proximal objects?

24. YES re: time. Each Planck-time creates a new iteration of the underlying data.

25. •X ( z q() Z ()Q z ) Y•

26. The smartest man of his generation. Thank you! Great presentation.

27. This guy is a true hero. Gets a ring from the muse – engages the chaos – kills the dragon single handedly – and brings gifts of wisdom back down to Platos cave. Most are unready and ungrateful. What makes him a hero is having the guts to do it anyway, dragging the ignorant kicking and screaming toward the finish line. Much love and gratitude Stephen. Thank you for all your hard work and generosit.

28. "AI" can only fix a "quantity" problem, not a "quality" problem. It can't "add new value", just regurgitate what we've already produced. If there's a fundamental fault in our understanding of existence, the AI will not bridge it, it will spend the world's power on the infinite generative recursion it takes to try and make broken science stick.

29. 00:10 Can AI really solve all scientific questions?
    02:34 AI in science focuses on machine learning and neural networks trained from examples
    06:59 Science faces computational irreducibility limits
    09:17 Computational irreducibility is a key challenge in science and math.
    14:01 AI and neural networks can be used to construct scientific models from data.
    16:20 Neural networks struggle with future predictions
    21:06 Diverse methods for time series prediction lead to varying success rates
    23:08 AI struggles to see beyond construction and training data
    27:24 Predicted probabilities evolve with successive rounds of training
    29:30 AI can potentially help in finding computational regularities
    33:30 Autoencoders can be successful in capturing the essence of images but may face challenges with scientific images due to computational irreducibility.
    35:37 Autoencoder compresses images into just two numbers
    39:28 AI can automatically discover patterns in complex data
    41:31 AI may capture raw natural processes different from human tasks
    45:20 Neural networks can find pockets of reducibility beyond human experience.
    47:14 Neural networks can capture some computational reducibility but not all.
    51:16 Using neural nets to solve the three body problem
    53:19 Neural Nets face computational irreducibility in solving complex systems.
    57:15 AI can help shortcut multicomp computational processes
    59:21 AI can help in generating sequences of tokens for mathematical proofs.
    1:03:16 Using neural networks to estimate distances and optimize paths
    1:05:11 Neural nets may lead us off track in finding paths efficiently.
    1:09:14 Exploring rules with computational irreducibility
    1:11:13 AI methods like reinforcement learning can help refine incremental steps in rule spaces
    1:15:41 Exploring different Paths of evolution towards living for exactly 50 steps
    1:17:43 AI's role in creating human-level scientific explanations
    1:21:36 Using machine learning to find formulas from data
    1:23:49 Using modern techniques to find a scientific narrative for data
    1:27:50 Discovering new science requires a flexible scientific narrative.
    1:29:48 AI and neural networks can learn to identify anomalies and surprises.
    1:33:42 Not all alkanes with 10 carbons are listed in knowledge bases.
    1:35:37 AI can potentially determine the relevance and significance of scientific findings
    1:39:32 AIS face the challenge of human choice and unpredictability.
    1:41:44 AI can help identify gaps in existing academic literature
    1:45:35 AI may struggle to create fresh conceptual frameworks in science
    1:47:32 AI can enable more definite and quantitative theories in non-formal sciences.
    1:51:26 AI measurements as a source of formalizable material
    1:53:21 AI can extend the domain of exact sciences.
    1:57:14 AI leveraging computational reducibility is a new tool for science
    1:59:19 Computational language accelerates science
    2:03:08 AI can mine patterns across diverse disciplines
    2:05:16 Encouraging interdisciplinary approach through analogies in education
    2:09:10 Time is the progression of computation in the universe and in our brains.
    2:11:19 Language as a modular structure and computational irreducibility
    2:15:31 AI can reliably create valid theories with limited information
    2:17:26 Using AI to automate identification of interesting objects
    2:21:48 Neural nets may not effectively learn from non-humanlike data
    2:23:53 Training data generated from AI may help in weights optimization but not in discovering new things
    2:28:11 AI can explore multiple parallel sequences of words unlike human brains.
    2:30:03 Exploring the essence of human consciousness and AI experience
    Crafted by Merlin AI.

30. Corporation ai solve science is a Turing type question. Pathetic, kind of.

Comments are closed.