Math notation is essentially a very compressed form of algorithmic/logical reasoning. If you’ve been doing this stuff it becomes the second nature to you, just like being able to go through a complicated codebase is for an advanced SWE with many years of experience. You are only
RESEARCH
-
LLMs lack memory consolidation like coworkers with anterograde amnesia
By
–
Agree that this is an important capability hole right now (I saw you push back on it a few times in the pod and I also didn't find the answers too satisfying). I like to talk explain it as LLMs are a bit like a coworker with Anterograde amnesia – they don't consolidate or build
-
LLMs Tax Returns vs ASI Development Priorities
By
–
Dwarkesh asking Sholto whether LLMs can do tax returns in the next couple of years and essentially getting nah we are too busy polishing the ASI
-
LeCun’s AGI Skepticism: Why Scaling LLMs Alone Won’t Suffice
By
–
When I hear LeCun say:
— Marek Rosa | European🇪🇺 | South African🇿🇦 (@marek_rosa) 3 juin 2025
"We are not going to get to human-level AI by just scaling up LLMs. This is just not going to happen. There's no way", the only thing that comes to mind is this quote:
"The cynics get to be right and the optimists get to be rich. 99% right, and 100% wrong.… https://t.co/skpLkli88SWhen I hear LeCun say:
"We are not going to get to human-level AI by just scaling up LLMs. This is just not going to happen. There's no way", the only thing that comes to mind is this quote: "The cynics get to be right and the optimists get to be rich. 99% right, and 100% wrong. -
The Inverse 80-20 Rule in AI Research: When Final Effort Matters Most
By
–
The 80-20 rule happens often in AI research, where you get 80% of the payoff from the first 20% of the effort. But there is often also an inverse rule, where it’s actually the final 20% of that yields 80% of the payout. Some examples: 1. When your eval is already good in many ways but it has one drawback and fixing that will extend its life from 6 months to 2 years – When you’ve already done all the experiments and writing them up in a nice report will allow dozens of people to learn from the work you already did – When you’ve spent some time debugging and you have a workaround. Instead of moving forward you spent the time to fully understand it; this knowledge compounds – When you’ve already run four experiments to investigate how much a particular component matters in training, and running one more allows you to find the answer (understanding things more clearly also compounds) Key is to know when it’s a 20-80 and stop and when to do the whole thing
→ View original post on X — @_jasonwei, 2025-06-03 19:40 UTC
-
Functions for Loading, Storing, and Manipulating Information Systems
By
–
i guess the others are used to represent functions which load, store, and otherwise manipulate information
-
Resource Gap Between AI Goals and Implementation Reality
By
–
The difference between “ought” and “is” is usually a couple of order of magnitude in terms of resources. Oftentimes much more than that.
-
Competitive Intelligence Transforms with AI Integration
By
–
Why competitive intelligence needs AI now more than ever https://
linkedin.com/pulse/why-comp
etitive-intelligence-needs-ai-now-more-than-ever-babin-b2pbe
…
#marketing #artificialintelligence #Competition #Innovation #technology @BetaMoroney @CurieuxExplorer @Shi4Tech @Khulood_Almani @Hana_ElSayyed @PawlowskiMario @JolaBurnett @lyakovet @enilev @AkwyZ -
Toolformer: Empowering Language Models with Tool Integration
By
–
Welcome! Toolformer is one my favorite recent papers.