thewarrior

You’re absolutely right !

While I agree with the points he’s raising let me play devils advocate.

There’s a lot more code being written now that’s not counted in these statistics. A friend of mine vibe coded a writing tool for himself entirely using Gemini canvas.

I regularly vibe code little analyses or scripts in ChatGPT which would have required writing code earlier.

None of these are counted in these statistics.

And yes AI isn’t quite good enough to super charge app creation end to end. Claude has only been good for a few months. That’s hardly enough time for adoption !

This would be like analysing the impact of languages like Perl or Python on software 3 months after their release.

Nope this is a preview of their coming agentic product. Could be something like a browser + deep research + computer use agent for consumer use cases.

Mods - earlier submission had a rendering bug and I had to create a new version hence the resubmission.

Any chance that “reasoning” can fix this

I use generative AI every day.

Atomic energy has both good and bad uses. People being tired of atomic energy has held back GDP growth and is literally deindustrialising Germany.

AI isn’t 3D TV

That’s always the case with any new technology. Tech isn’t going to make everyone happy or achieve world peace.

It’s writing most of my code now. Even if it’s existing code you can feed in the 1-2 files in question and iterate on them. Works quite well as long as you break it down a bit.

It’s not gas lighting the latest versions of GPT, Claude, Lama have gotten quite good

I’m tired of farming - Someone in 5000 BC

I’m tired of electricity - Someone in 1905

I’m tired of consumer apps - Someone in 2020

The revolution will happen regardless. If you participate you can shape it in the direction you believe in.

AI is the most innovative thing to happen in software in a long time.

And personally AI is FUN. It sparks joy to code using AI. I don’t need anyone else’s opinion I’m having a blast. It’s a bit like rails for me in that sense.

This is HACKER news. We do things because it’s fun.

I can tackle problems outside of my comfort zone and make it happen.

If all you want to do is ship more 2020s era B2B SaaS till kingdom come no one is stopping you :P

It’s not reproducing exact strings in the training data but patterns and patterns of patterns.

Next token prediction is more intelligent than it sounds

What are some interesting kinds of graphs one can build this way and use them ?

There’s a giant caveat here - this assumes that the current LLM architecture is enough to bootstrap to those higher levels of intelligence. LLMs are incapable of some pretty simple things at this point and it’s a big question mark of whether they are even capable of doing sophisticated reasoning and planning architecturally.

GPT-4 cannot play a good game of tic tac toe. But it can play passable chess. This is a good point to ponder.

“Rewrite this sentence to be more concise”

This is where speech to text models like Whisper and LLMs can make a real difference.

I have friends in academia who use GPT-4 to help with with research level code. TikTok just released an app where you can hum a song it will generate a full instrumental backing track.

This stuff can already do impressive things and its only getting better.

I would urge everyone to hold back their instinctive reaction to the usual SV hype and go and try GPT-4,Claude+, Mid Journey, RunwayML for a few weeks and come to their own conclusions.

At a high level yes.

More precisely - It gets the question After irs passed through a matrix that transforms the text description of the image so the LLM can “understand” it.

It maps from the space of one ML model to the other.

I've only been reading ML stuff for a few months and I kind of understand what it's saying. This stuff isn't as complex as its made out to be.

It's just a bunch of black boxes AKA "pure functions".

BLIP2's ViT-L+Q-former AKA

    //I give you a picture of a plate of lobster it will say "A plate of lobster".

    getTextFromImage(image) -> Text

Vicuna-13B AKA

    //I give you a prompt and you return completion ChatGPT style
     getCompletionFromPrompt(text) -> Text

We want to take the output of the first one and then feed in a prompt to the LLM (Vicuna) that will help answer a question about the image. However the datatypes don't match. Lets add in a mapper.

    getAnswerToQuestion(image, question) -> answer 
        text = getTextFromImage(image)
        prompt = mapTextToPrompt(text)
        return getCompletionForPrompt(prompt)

Now where did this mapTextToPrompt come from ?

This is the magic of ML. We can just "learn" this function from data. And they plugged in a "simple" layer and learned it from a few examples of (image , question) -> answer. This is what frameworks like Keras, Pytorch allow you to do. You can wire up these black boxes with some intermediate layers and pass in a bunch of data and voila you have a new model. This is called differentiable programming.

The thing is you don't need to convert to text and then map back into numbers to feed into the LLM. You skip that and use the numbers it outputs and multiply directly with an intermediate matrix.

    getAnswerToQuestion(image, question) -> answer 
        text = getEmbeddingFromImage(image)
        embedding = mapEmbeddingToInputEmbeddingForLLM(text)
        return getCompletionForEmbedding(embedding)

Congratulations you now understood that sentence.

We’ve tried but we understand even less about how “real” intelligence works. Atleast in a way that we can replicate it.

The people doing so called alchemy are making far better systems that anyone who’s trying to actually understand things. And we’ve been trying for decades now.

Had GPT-4 clarify some of the syntax for me https://poe.com/s/uj80au2i4TuNWgUa1pNm

Does anyone have a TL;DR version of what’s going on here

Have you tried getting it to explain the code to you

How long did it take you ? This tech turns days of work into hours or minutes of work.

How often do you think we should post if we have made non trivial improvements

Iterating based on HN Feedback :).