In the past, most ML research was driven by academia and there was a circle from people at UoToronto, UdM, Stanford, Berkeley etc. Almost everything that came out or even most pioneers were here. Now academia isn't dead but lost it's place.

Now the circle is OpenAI, Anthropic, Meta, Deepmind, XAI, Microsoft Research etc. Most of the people at the academia circle in the above are running these labs now but with less freedom to explore ideas over time.

Here comes the downfall in the second circle, in academia people spend 5 years to make some cool things and they did, most of the things were initiated in uni labs than industry. But now everyone is chasing benchmarks and 1% gain over the SOTA is almost enough to lead the race. Ofc industry made so many great things, but the same way scaling is converging, it'll be for industry labs, unless they have teams to do foundational research that might take years, e.g FAIR(Meta).

rwanda and those cheerful ladies are becoming more and more attractive

Is this really true?

Well, one way to find out

How many of you like or work on Computer Vision? Are there like any good Ethiopian based researchers working on CV that do some cool things, just wondering.

Currently in Kigali at ACVSS, it'll be very good to join it next year if you like Computer Vision.

Better Auth was live in Times Square NYC last night. Crazy how far and how quick things go ❤️

Alphafold isn't that novel work. I just learned yesterday that there were people who almost showed this 3 years prior and published a paper at a top conference, Neurips. Alphafold did almost exactly the same, used deep networks for prediction and scales it. However, they didn't even cite their paper, whyyyy sus

I learned that from Daniel Cremers yesterday and they were the ones who deserved the nobel prize.

So many NP-complete problems out there. What a time to be alive and see them solved/solving, proven, or just discovering them.

Machine learning is just statistics. On steroids. Lots and lots of steroids.


X

Machine learning is life. Karaoke is also life. Therefore machine learning is karaoke

X

Somewhere in Kigali

For anyone wondering about this, there are many who look Ethiopian, that's all I can say! Answer found so time to get back✌️

Anthropic🤔

Judging in a hackathon is really hard!

Key take aways from the hackathon:

- People trust AI too much and not do much of customizations.

- Super far fetched ideas that are very idealistic.

- You can build great projects with Vibe coding these days, which is cool.

- You don't need to solve big problems, you can just build fun/cool things for a hackathon.

Maybe don't use an LLM for _everything_?

They showed that agentic pipelines that mix LLM-prompt steps with principled techniques can yield better, more personalized summaries

https://arxiv.org/abs/2309.09369

This is really good. I finished watching it and reading the paper.

ICML 2024 Tutorial: Physics of Language Models

https://www.youtube.com/watch?v=yBL7J0kgldU

Seems a bit better than last year for us. You could count the number of Ethiopians using one hand last year. In fact we were only 3 I'd say.

ACL 2025

This is a cool list. Alec and Noam being in top 5 with no Phd is the greatest thing ever.


https://www.metislist.com/

I'm glad to be ranked 80th world wide, thanks Francois Chollet.