@oxfordmathematics: A computer scientist walks into a mathematics department... AI is in Frankenstein mode, worrying about the monster it has created. Even the tech bros seem terrified. So, where does mathematics fit in to AI's future? #maths #ai #machinelearning #stem

Latchezar Tomov :

Is computer science really separate from maths?

Brian :

AI is solely done for a profit gain for businesses. AI is just darkness waiting to be seen. As for us average people we don't need it much other than just explaining stuff in simple terms.

iceman.sid :

PhD from uoft oh this guy is smart smart

user4855775220723 :

Naive and forgot philosophy

NotMaxFreedom :

Interesting take. I think math is a great check on AI integrity as well. Human input can "naturally" boost why is being branded as "artificial" intelligence. This helps in the short term but destroys long term prospects. One reason I am confident that Elon Musk's Grok from xAI will eventually be the best and most powerful AI, is that despite being much weaker currently, it scores 100% in mathematics - far above more powerful and higher ranking AI like the latest from ChatGPT and Claude. This tells me that the AI is being built with much less errors, even if that means slower initial growth. As Benjamin Franklin once said in his autobiography, which I recommend everyone read: "I doubted whether some error had not insinuated itself unperceiv'd into my argument, so as to infect all that follow'd, as is common in metaphysical reasonings." The linear and additive nature of logic means the logical distance/height that can be reached by intelligence has a hard limit put on it by early misconceptions, since intelligence is just the combination of knowledge points. Two data points is knowledge, two data points being combined into a third is intelligence. Therefore the integrity of early data points sets the final limit. Mathematics is the best check for integrity that I can think of, and to finally get to the point, I disagree that any limit has been hit by AI. We're hitting a natural limit based upon natural factors, and you are suggesting we fix them in a "natural" way - with more people. Wrong mentality for AI

Kena Dams :

What is his name or research lab?

goodmorningsteve :

“Platoood”

Mythrra :

Hallucinations ?

Prospero Prime :

future of AI is math 😎😎😎😎 check out axiom

_Aeshma_ :

Can 2 fold logics handle future of AI ?

Kev :

I heard what he said, but what is he really saying? 😳😳😳😳

Antonio G Bucio :

The first computer scientists were mathematicians. Many CS departments come from the math department in universities around the world. Type Theory is a mathematical theory of mathematics and computation.

SciencePen :

LLMs now (Claude, ChatGPT, ...etc) = Linear Algebra, probability, multivariable calculus. These are basic Mathematics, he said what if we use (or explore) more advanced math.

mikak :

he.talks!

Trufflepesto and 307 others :

Will come from biology 👍

Kevin 🇨🇦 :

The books on the shelf behind him…..🫣

Anushka Lawrence :

Lots of words to say future development of ai needs to be targeted.

Kimochi69 :

KKKKK :

First computer scientist were mathematicans

CHZN1 :

I regret not taking more math in high school although the teachers weren’t great. Now I’m learning Linear Algebra and the hardest part is all the notation. Calculus looks even stranger lol

polarbear1974 :

Yep roots laid down now trunk branches ...flowers fruit and seeds to follow. leaves with phothosynthesis to sustain.

Steven oc :

Sounds to me that he's saying, we don't know. Effectively, we don't know what we don't know. So we're just going to keep on experimenting until we see something we do know. But how will we know what we don't know? I'd really like to know how that's possible. Maybe we've already reached the outer limits of what's usable knowledge. How will we know?🤪🤪🤪😎✌️

Fisi :

Just to say mechanistic interpretability.