gelisam

@haskell_cat

AI Safety ∩ Programming Language Theory. Part-time technical alignment researcher, full-time Haskell software engineer at http://well.co, opinions are my own.

Montreal

haskellcat.com

Joined June 2016

2KPosts 1KFollowers 313Following

You might like

@bgamari

@tweagio

@welltyped

@kosmikus

@mattoflambda

@nomeata

@snoyberg

@int_index

@tikhonjelvis

@vrom911

@csaba_hruska

@Iceland_jack

@owickstrom

Pinned

gelisam

@haskell_cat

Oct 9

I am doing technical alignment research in my free time. Here is a project in which I use static analysis to verify whether a neural network satisfies its safety property under _all_ inputs or if it needs more training. gelisam.com/ai-safety-via-…

haskell_cat's tweet card. Proving that the neural network learned its safety property.

Parity Bot

Source: gelisam.com

gelisam

@haskell_cat

Dec 8

Tell your model _why_ ask! "Would you like me to explain how to downgrade your compiler, port the code to a newer version... or, since you want to use this code as a starting point, would you like me to generate some starting code which works with your version of the compiler?"

gelisam reposted

Joe

@alpha_convert

Nov 25

I know I'm a little late to the unwrap discourse, but: The value proposition of Rust is not "your program will not crash." It's "your program will not have memory unsafe behavior" The widespread conflation of the latter with the former is due to Haskell being too successful :)

gelisam

@haskell_cat

Nov 6

We should not develop superintelligence until we know how to do it safely. Sign the statement today! #KeepTheFutureHuman superintelligence-statement.org

haskell_cat's tweet card. “We call for a prohibition on the development of superintelligence, not lifted before there is (1) broad scientific consensus that it will be done safely and controllably, and (2) strong public...

Statement on Superintelligence

Source: superintelligence-statement.org

gelisam reposted

ControlAI

@ControlAI

Oct 22

🚨BREAKING: A tremendous coalition of leaders just called for a ban on the development of superintelligence. The two most-cited living scientists, Nobel laureates, faith leaders, politicians, founders, artists, a former head of state, and over 1000 others have joined. Thread 🧵

ControlAI's tweet image. 🚨BREAKING: A tremendous coalition of leaders just called for a ban on the development of superintelligence.

The two most-cited living scientists, Nobel laureates, faith leaders, politicians, founders, artists, a former head of state, and over 1000 others have joined.

Thread 🧵

gelisam

@haskell_cat

Sep 26

Somebody on my timeline recommends a book and it is _not_ If Anyone Builds It Everyone Dies??

Santiago

@svpino

Sep 25

This is a really good book. I like it because it covers both ends of the spectrum: 1. How LLMs work 2. How to build using LLMs It's a really nice one-two punch: start with the theory and use that right away to implement something useful. The second half of the book is what I…

svpino's tweet image. This is a really good book.

I like it because it covers both ends of the spectrum:

1. How LLMs work
2. How to build using LLMs

It's a really nice one-two punch: start with the theory and use that right away to implement something useful.

The second half of the book is what I…

gelisam reposted

Jeffrey Ladish

@JeffLadish

Sep 24

Post: lesswrong.com/posts/6ZnznCaT…

gelisam

@haskell_cat

Sep 4

> my prediction is that auto-regressive LLMs are doomed Yan LeCun, notorious AI Doomer 🤣

Karl Mehta

@karlmehta

Sep 3

1. "Nobody in their right mind will use autoregressive LLMs a few years from now." The technology powering ChatGPT and GPT-4? Dead within years. The problem isn't fixable with more data or compute. It's architectural. Here's where it gets interesting...

gelisam reposted

Crémieux

@cremieuxrecueil

Aug 16

One concept I wish more people were aware of is the Tocqueville Effect. Named for Alexis de Tocqueville, this concept describes the curious phenomenon by which people become more frustrated as problems are resolved: As life gets better, people think it's getting worse!🧵

cremieuxrecueil's tweet image. One concept I wish more people were aware of is the Tocqueville Effect.

Named for Alexis de Tocqueville, this concept describes the curious phenomenon by which people become more frustrated as problems are resolved:

As life gets better, people think it's getting worse!🧵

gelisam reposted

Rob Miles

@robertskmiles

Aug 22

Technology is generally really good. Why should AI be any different? A new video: (youtube link in the reply)

gelisam

@haskell_cat

Jun 26

When I was a teen, my government held a public consultation about switching from 1st-past-the-post to a proportional system. I did my research, went, presented approval voting, and was laughed off the stage. I never did politics again. Today, we're still using 1st-past-the-post.

Nauseam

@ChadNauseam

Jun 25

If you liked this, follow @ElectionScience for more information on a better way! It's called "approval voting", and it's so simple: you can vote for multiple candidates, and the candidate with the most votes wins. While not perfect, I think it's better than ranked-choice

gelisam

@haskell_cat

Jun 20

I like to learn about neural networks by working on tiny problems for which there exists a 100% correct solution. Here is an interactive experiment showing how in practice, backprop doesn't find this solution: gelisam.com/local-minima/

haskell_cat's tweet card. Tinker with a real neural network right here in your browser.

Tensorflow — Neural Network Playground

Source: gelisam.com

gelisam

@haskell_cat

May 30

My first MCP server, which allows the agent to pick from a selection of shell commands. GitHub Copilot can natively run shell commands, but VS Code asks you to confirm each command. With mcp-cli, you only have to authorize the use of the tool once! github.com/gelisam/mcp-cli