Non-influencer (a feature, not a bug) | Investor, AI strategist, amateur number theorist | 𝝋 Omnia numeri sunt 𝝋
Cape Town, South Africa
Joined September 2008
- Tweets6.1K
- Following209
- Followers369
- Likes6.7K
Not ideal that Exponential Time Hypothesis has the acronym ETH, especially when there is speculation that OAI has broken it...
$ETH
Yikes. Things are moving fast 🙈
"Major U.S. banks are detecting unauthorized third-party AI agents operating inside customer accounts, with some agents attempting to take actions that appear to be independent of user instructions, sources tell Axios." axios.com/2026/10/22/ai-agen…
"Demand vs. Supply" obfuscates something very important.
The accurate rendition:
"Profitable Demand versus Profitable Supply."
...and that is much more difficult to measure.
Mechanistic Interperabilty is plausible.
The argument that because we don't build AI, but grow AI, we are prevented from understanding its inner workings, is short sighted.
A farmer that threw a seed in the ground 4 centuries ago and saw the plant grow, knew nothing of the inner workings of plant biology either...
A common go-to analogy when comparing ourselves to a superior AI intelligence, is that it will be similar to us and out pets.
A common blindspot on the other extreme is that it will be like us and our livestock.
I'm thinking the reality will be in-between: AI will treat us like zoo animals. And once you entertain that possibility, then you have to accept the possibility that we've been observed as zoo animals for more than a decade.
I am a bit torn apart about Terence Tao's new mathematical guild. I am a lover of math, but as an amateur number theorist I have experienced the "stay out of our lane" hostility from the pro math community. So on the one hand I find it ironic that they've directed their "stay out of our lane" anger for the first time in the opposite direction from the weaker position.
On the other hand all scientific progress is built on the back of mathematics, perhaps AI more than any other. So I get the indignancy of "our creation has turned in us."
Sadly, I think this moment in time is remiscent of the Pythagorean's outlawing of irrational fractions. Futile. But in terms of respect I also think that the Pythagoreans are, to this day, the guild most revered by modern mathematicians.
"Computers will have superhuman smell. Sensors will be able to smell disease on people and crops. It will be possible to monitor industrial processes more effectively as well. Many strange implications here, but potentially exciting for biosecurity."
Fascinating.
Ok two more:
1. AI will enable the design of custom enzymes that can catalyze industrial processes and chemical manufacturing faster than existing methods. In some cases this will make existing chemicals cheap enough that it becomes possible to use them for totally new use cases. Also, custom chemicals may enable the design of more efficient molecules to do things like separate rare earths from other elements, currently a very nasty process.
2. Computers will have superhuman smell. Sensors will be able to smell disease on people and crops. It will be possible to monitor industrial processes more effectively as well. Many strange implications here, but potentially exciting for biosecurity.
In the history of homo sapiens, we lot, here and now, are very unique in our positioning on the long scale timeline.
"I feel like the small mammal that I am, awed and afraid."
An articulate response to OAI's monster breakthoughs in their latest math drop.
Incoming: many small mammals, soon.
This #koan hits different in today's AI world:
"If we were clever enough to understand our own brains, we'd be so stupid that we wouldn't be able to understand."
I'm getting serious cliffhanger fatigue.
All the highstakes things in the world in so many domains have seemed to be teetering on going seriously wrong for so long now - its just exhausting.
Think I must sign up to the "nothing ever happens" camp for a while.
"AI first blazes the trail, and humans decipher it to extract meaning." 🎯
5000年の数学史で初めて、「理解」と「発見」が分離し始めている。AIが人間には理解が追いつかないほど大量の正しい数学的発見を生み出す世界で、「数学をする」とは何を意味するのだろうか。数学は「まずAIが道を開き、人間がそれを解読して意味を回収する学問」に変わってしまうかもしれない。
サンヒョン・キム「私たちは今、大きな混乱と変化のただ中にいます。数学の5000年の歴史でも経験したことのない変化です。
数学は、他の科学分野とはかなり異なっていました。トーマス・クーンの科学革命論でも、数学はその枠組みには当てはまりませんでした。数学ではずっと同じパラダイムが続いていて、偉大な定理を証明する、つまり大きな発見をし、それを深く理解することが目標でした。それが5000年間の数学の歴史だったのです。
最初は役に立たないように見えても、数百年、あるいは数千年後には人類にとって極めて有用になることもありました。しかし歴史上初めて、この二つが分離し始めています。「理解」と「発見」です。
これまでは両者は一致していました。大きな発見をするということは、その対象を非常によく理解していることを意味しており、それこそが数学にとって望ましいことでした。
ところが今、私たちはAIが生み出す発見に圧倒されています。もちろん間違いもありますが、そうした間違いはこれからますます減っていくでしょう。AIはさらに進歩し、その答えや発見もますます正確になっていきます。
そして数学者たちは気づき始めています。私たちはあまりにも多くの発見を手にする一方で、それに対する理解ははるかに少ないのです。今まさに、それが起きています。私たちはこの状況を非常に懸念しています」
Aggressively rewarding LLMs for WHAT they acheive but not HOW it is acheived is great for benchmark progress. Have a guess how good it is for alignment...
#alignment
So if the US is approaching AI escape velocity with AI designing AI,
is China approachinh Robot Manufacturing escape velocity with robots manufacturing robots?
#escapevelocity