Tao is out of his lane; lots of medicines don't have understood mechanisms. We don't even have the mechanisms behind general anesthesia nailed down; do you want to forgo it when the docs cut you open to remove your tumors?
Now math gets to deal with that same reckoning. They were already well on their way there with previous Lean proofs, but this has pushed things beyond that horizon and I'm not sure some of the mathematicians are ready for it.
As a comparison, classical computing has been scaled up a bazillion fold too, and can do things which are absolutely miraculous, but every layer of abstraction is discretely understandable.
Also, while biological systems simply exist in nature, artificial neural networks are ultimately mathematical objects with various properties that have yet to be uncovered.
But just to highlight the issue with his statement: if I have cancer and I'm dying of cancer and there's a promising treatment which hasn't been fully understood by even a single human researcher but which has demonstrated clear positive outcomes, then I'm gonna take it. I agree that this is really just a balancing of risks, so I likely wouldn't take such a cocktail if it was just to, say, cure my sniffles. But when the alternative is terminal illness, you'll find that people become willing to accept absolute quackery for cures, so an AI produced cure that, again, has enough evidence to make me believe it is worth a shot is not a stretch.
Terence hasn't won a Nobel, but this reminds me of "Nobel disease":
"Nobel disease or Nobelitis is an informal term for the embrace of strange or scientifically unsound ideas by some Nobel Prize winners, usually later in life. It has been argued that the effect results, in part, from a tendency for Nobel winners to feel empowered by the award to speak on topics outside their specific area of expertise, although it is unknown whether Nobel Prize winners are more prone to this tendency than other individuals."
I often used to tell my students that (plausible) mechanism is what distinguishes productive science. For example, population genetics was on very solid empirical and theoretical grounds but until we understood how DNA replication / transcription worked from 1953 onwards, genetics had a sort of unsatisfying magical, fairy-tale feel about it.
This is contrary to what some philosophers of science will tell you about measuring progress science purely in terms of its predictive power.
What I have to say applies really only to my fields of medicine and biology. I have no idea how physics or chemistry works as practical endeavours.
It's very easy to say that it shouldn't have to be the case with the hindsight of GA being successfully used for centuries.
I would be much more interested in an example of a drug/medical procedure/research that currently is having to jump through such hurdles that people think should be treated as permissively as GA or the theoretical 'cancer cure', despite there not yet being any evidence of its long term effects - as would be the case with a hypothetical AI 'cancer cure' that's not understood by medical professionals.
Personally I can see the arguments for both sides of the coin, but I find GA to be a bad example.
The LLM model was able to break down and express math in such a simple way that I could understand and follow the training of the LLM model itself!
Are the math and computation accurate? I don’t know, and likely there are significant errors. Nonetheless, if I had a little more time -- not infinite time -- I would be able to prove whether they are or not.
Likely this is the path forward for understanding the mechanisms of medicine, and since most humans learn by doing and interacting with an environment, interacting like this will become how we use AI for learning in the near future.
You are bringing up a good point. Stupid me can be presented with on demand pretty animated cartoons in order to understand anything known. Imagine how this is going to improve learning speed the student on the engineering track? In a couple years, most engineering students will learn this in the first year of undergrad!
Would you prefer to take the medicine that is proven to work or the one that is quite interesting for academic reasons behind its understood mechanisms but doesn't actually work?
You run an experiment to test your hypothesis. It's not like you can build a model with zero data. You can only make predictions at best.
You guys act like understanding something requires complete and global knowledge about everything.
> Suddenly, with this AI math thing in the past year or two, he is starting to get more non-academic focused attention.
He's just gaining popularity in your bubble, which is probably not much larger than the audience he previously had. It's not like he's suddenly become orders of magnitude more famous.
What I would expect of someone who goes to an lecture of a professor. I never said he was completely unknown; just that he didn't have widespread recognition outside of academia. Getting autographs after an academic lecture doesn't dispute this.
>He's just gaining popularity in your bubble, which is probably not much larger than the audience he previously had.
The reason why he is gaining this popularity isn't anything to do with his academic achievements though. It's because of him being a child prodigy that is commenting on AI in math. So yes, he is reaching much further than his base, especially because of OpenAI's advancements in math and what it means about AI capability going forward. He has essentially positioned himself as the voice of mathematicians' position on AI and math.
You (NewsaHackO) made a claim, which the other commenter showed was wrong. Then your successive comments change the nature of the claim / qualify your statement to a point it is a different one.
This is just an observation, It is not my place to make a judgment. We all learn by nuancing our understanding, so the thread may have widened everyone's worldview. I learned something and thank you both for this tidbit.
But it is also important to be humble: we are all wrong about many things. Most of those things don't matter to us as a person or to our lives. Thus when there is a (presumably) low-importance opinion as the one discussed here, maybe it is a good moment to take a step back, reflect, and update our worldview.
It helps us grow when we update our beliefs, and it is a sign of a wise man that he can accept that he was wrong and learn from it, rather than to argue about semantics or reframe the debate.
Reducing him to child prodigy is, I don't know, condescending? You do realise he's one of the greatest living mathematicians, right? Child prodigy sounds as if his greatest life achievements was being smart kid or something, not his substantial contributions to mathematics as an adult. He was one of the most important voices in maths long before transformers were discovered.