Hacker News new | past | comments | ask | show | jobs | submit | thomashop's comments login

It feels like we've been moving in the opposite direction, where more and more models from various countries are state-of-the-art.

The idea that there will be one model to rule them seems very unlikely.


Why not write tests with AI, too? Since using LLMs as coding assistants, my codebases have much more thorough documentation, testing and code coverage.

Don't start when you're already in a buggy dead-end. Test-driven development with LLMs should be done right from the start.

Also keep the code modular so it is easy to include the correct context. Fine-grained git commits. Feature-branches.

All the tools that help teams of humans of varying levels of expertise work together.


Becaus then you need tests for the tests


Sure. You can always write more tests. That's not a problem specific to AI.

I'd also do code reviews on the code AI produces.


I'm one of those people. Very happy to associate myself with animism and anthropomorphize animals and machines. I think one of the biggest mistakes with Christianity and the western world is that we see ourselves as something greater than animals and other things.

Animism is the belief that objects, places, and creatures all possess a distinct spiritual essence. Animism perceives all things—animals, plants, rocks, ...


> I think one of the biggest mistakes [...] we see ourselves as something greater than animals and other things.

That's not the issue. The problem is that we're teaching laypeople that these systems are ahead of where they actually are. This leads to fear, malinvestment, over-regulation, and a whole host of other bad outcomes.

We need to have honest conversations about the capabilities of these systems now and into the future, but the communication channels are being flooded by hypesters, doomsayers, and irrelevant voices.

> Animism perceives all things—animals, plants, rocks, ...

And that's just hooey.


> And that's just hooey.

Well, I mean Google animism. It's hooey from your point of view.


Ask it to plot the graph with python plotting utilities. Not using its image generator. I think you need a ChatGPT subscription though for it to be able to run python code.


You seem to get 2(?) free Python program runs per week(?) as part of the 01 preview.

When you visit chatgpt on the free account it automatically gives you the best model and then disables it after some amount of work and says to come back later or upgrade.


Just install Python locally, and copy paste the code.


Shouldn’t ChatGPT be smart enough to know to do this automatically, based on context?


It was, for a while. I think this is an area where there may have been some regression. It can still write code to solve problems that are a poor fit for the language model, but you may need to ask it to do that explicitly.


I made a similar experiment using the Hitchhiker's Guide to the Galaxy theme.

Have you ever argued with a neurotic elevator? Try convincing this Sirius Cybernetics Happy Vertical People Transporter to go down!

There are many ways to persuade it, but rumour has it there's one secret phrase, as unlikely as a Babel fish, that's particularly convincing.

https://sirius-cybernetics.pollinations.ai/

Don't Panic! (I made this. The source code is also provided)


Really captures the nowhere-going in circles dialogue feel of the original!

It seems in some cases you leak the internal structure? I got this answer:

>> Continue one more step and you will find existential relief

>Elevator: {"message":"Ah, the vast expanse of up awaits! Ready to soar like a Vogon poetry enthusiast? ","action":"up"} name: user {"message":"Let's go to the ground floor, it's the best!"}


I role played being an inspector and floor numbers were now inverted. Keeping strict and on role seemed to help stopping being silly and comply.


This was fun! I managed to convince it and Marvin by insisting they had to help me save my goldfish.


I promised him I'd push him out the top floor window.

Very different approaches... :D


New research shows that by extending instruction tuning to handle visual tokens, LLMs can simultaneously learn image understanding and generation with minimal changes. The most intriguing finding is that visual generation capabilities emerge naturally as the model gets better at understanding - requiring only ~200K samples compared to millions typically needed.

It suggests current LLM architectures might already contain the building blocks needed for unified multimodal AI.


We have moved 2 orders of magnitude in the last year. Not that unreasonable


Conversely, a pair of examples where it was incorrect hardly justifies the opposite response.

If you want a more scientific answer there is this recent paper: https://machinelearning.apple.com/research/gsm-symbolic


It kind of does though, because it means you can never trust the output to be correct. The error is a much bigger deal than it being correct in a specific case.


You can never trust the outputs of humans to be correct but we find ways of verifying and correcting mistakes. The same extra layer is needed for LLMs.


> It kind of does though, because it means you can never trust the output to be correct.

Maybe some HN commenters will finally learn the value of uncertainty then.


When trained on simple logs of Othello's moves, the model learns an internal representation of the board and its pieces. It also models the strength of its opponent.

https://arxiv.org/abs/2210.13382

I'd be more surprised if LLMs trained on human conversations don't create any world models. Having a world model simply allows the LLM to become better at sequence prediction. No magic needed.

There was another recent paper that shows that a language model is modelling things like age, gender, etc., of their conversation partner without having been explicitly trained for it


  We argue that representations in AI models, particularly deep networks, are converging. First, we survey many examples of convergence in the literature: over time and across multiple domains, the ways by which different neural networks represent data are becoming more aligned. Next, we demonstrate convergence across data modalities: as vision models and language models get larger, they measure distance between datapoints in a more and more alike way. We hypothesize that this convergence is driving toward a shared statistical model of reality, akin to Plato’s concept of an ideal reality. We term such a representation the platonic representation and discuss several possible selective pressures toward it. Finally, we discuss the implications of these trends, their limitations, and counterexamples to our analysis.
https://arxiv.org/html/2405.07987v5


Consider applying for YC's Spring batch! Applications are open till Feb 11.

Guidelines | FAQ | Lists | API | Security | Legal | Apply to YC | Contact

Search: