I was able to get gpt4 to do a lot of useful work. But for some reason it completely falls apart for this scenario. May be because it has to think in second order to achieve the task. Perhaps you could take a crack at this:

Prerequisite (for you the human)> You have a file at src/SampleReactComponent.jsx that has below simple react component: const SampleReactComponent = (props) => {

    const [var1, setVar1] = React.useState(false);
    const [var2, setVar2] = React.useState(false);
    const [var3, setVar3] = React.useState(false);

return (<></>); };

export default SampleReactComponent;

********** Prompt for GPT4: I'm at my project root working on a reactjs project. Update the component in src/SampleReactComponent.jsx file by adding a new const variable after the existing variables. You cannot use cat command as the file is too big. You can use grep with necessary flags and sed to achieve the task. I'll provide you the output of each command that you generate. *************

That's it. It would do any complex modification on fully provided data (included in the prompt) but something like above where it has to build a model from secondary prompts will totally fall apart.

Dude. Dude.

I’m an IT professional and I have no idea how to begin answering that request! Pose that question verbatim to a dozen random people[1] and I guarantee you that you’ll get zero answers.

Also, I find it hilarious that sed and awk are so counterintuitive that not even the AIs can do useful things with them. The same AIs that speak Latin, and can explain quantum mechanics.

[1] I mean specifically not random Silicon Valley coworkers. Go talk to random relatives and the barista at the cafe.

That would mean nothing because GPT4 isn't most people. I had it solve more complex problems than this particular one and using the same tools :)

Interesting! I, for one, would be most appreciative if we could see some of the similar and more difficult tasks that it succeeds on.

I have cleared my chat history recently so don't have all the prompts but here is a recent use case I used it for: Prompt: "Generate a react mui component that fetches a Seller's shipping credits and shows his current credit balance in the top portion. There should be a recharge button next to the credit balance. The bottom portion should show recent transactions. Use below api for fetching credit balance: <api req/resp>. Use below api for fetching the transactions: <api req/resp> Recharge should be done with the below api: <api req/resp>

Make the component beautiful with good spacing, elevation etc. Use Grid/Card components."

Final result: https://res.cloudinary.com/dksmi6x98/image/upload/v168527409...

Thank you for your reply. The impression I'm getting is that it is producing concrete solutions to specific problems, but not generic solutions to abstract problems - does that sound about right?

Kind of I guess. If it is a single step: That is modify or generate text (could be code or anything) in a specific way, it works great even on obscure things.

However if the problem is stated in a way that it has to think through derivative of it's solution: that is generate some code that generates some other code which behaves in a certain way, it fails miserably. I'm not sure why. The problem I stated in current thread which it failed on I have tried multiple prompts to make it understand the problem but unfortunately nothing worked. It's as if it can do first level but but not second level abstraction.

I'm assuming something like tree of thought needs used on problems like this. GPT 'mostly' thinks in a single step.

Also, in general we attempt to make these models as economically efficient as possible at this point because of the extreme lack of excess computing power to run these things. You can't have it spending the next 30 minutes thinking in loops about the problem at hand.

Thanks for suggesting tree of thought. Will try the approach they mentioned in the paper.

Give it space to think. It's like talking to someone and taking their train of thought as the only output, the worst kind of whiteboard programming interview.

You can get them to talk through the problem, build parts, test things, etc.

Also, this is a horrible problem to solve with awkward tools. Why is your code file to large to cat???

Most code files are usually too large to 'cat' because of the context size limitations. Even if they fit within the context window, it's waste of API credits to provide it the information that it doesn't need.

Anyways posting this here isn't to get this particular problem solved. It is to see if there is a prompt that can solve it. And this is the only problem I found it not able to solve. It's not like it doesn't know about sed/awk/grep or other Linux tools, it is an expert on most of the common options involving them. My guess is there is something going on with this prompt that just breaks it's 'though patterns' for the lack of a better word :)

> Most code files are usually too large to 'cat' because of the context size limitations

These are command line tools though, not piping it into the llm.

My point though was more about giving it space to reason. It's very important for good results on more complex tasks.

Aah, in case it wasn't clear the use case I'm trying to solve is exactly that: piping shell commands to and fro so that LLM has a bit of autonomy. I know that things like LangChain, AutoGPT exist but frankly they are really poorly thought out and seem to have become kitchen sinks too fast without solving anything properly.

Thank you for the pointer. Very interesting approach. If you don't mind could you suggest some resources where I could learn more about your approach.

Email your GitHub username to my HN username at gmail and I'll add you to the repo so you can take a look!

I think the issue is that it's trained with the assumption that it has all of the data it needs to answer. It's definitely tricky to get it to follow a data collection step and then stop before trying to complete a task. But it is possible. I think that langchain and ChatGPT demonstrates a good way to do this.

A lot of the people on the left of the curve are likely doing different shades of manual work. This is often work that automation is struggling to replicate --think grasping and packing in a warehouse, or tidying up a messy shelf of goods-- and the GPTs won't change this.

I'd argue it's maybe the middle and even the right of the curve that's threatened. Is your job writing marketing copy, or communications materials or press releases, or relatively simple moulding and counting of data, or HR administration (or many other examples)? GPT's coming for your job in a short space of time.

(Which brings us to mft's second rule: if your job is threatened by the current generation of GPTs, it should probably have been largely automated by simpler means --or just eliminated-- already.)

> and the GPTs won't change this.

I suspect but cannot prove that the opposite will be the case.

We can make bipedal robots now, but we can’t give them instructions, not like we can instruct human labourers.

“Bob, go to where George ate his lunch an hour ago and bring me the tools for fixing a pipe” was an impossible instruction for a robot to execute.

Suddenly this is a trivial sentence for an AI that knows what type of tools are typically needed for fixing pipes!

Think about how many mega projects are being built in places like Saudi Arabia with immigrant workers that speak little English and no Arabic. An AI is already superior in terms of comprehension than those workers!

LLMs already speak every major language and can be a handy translator too. C3P0 and R2D2 are looking totally realistic suddenly! Heck, do you remember that scene in the Empire Strikes Back where Luke used a text chat interface to communicate with R2D2? That’s oddly prescient now!

That's not quite the issue I'm raising. It's not about being able to communicate better with existing robots - it's where (as yet) there are no robots suitable to do the job.

There are plenty of tasks which are simple, tedious and repetitive - and yet are resistant (currently) to automation. An example is sorting mixed plastic waste into categories according to the type of plastic. This is comparatively trivial for almost any human, and (afaik) not yet possible to automate. (Another is picking and packing things in Amazon's warehouse - which we know they've expended huge effort in trying to solve.)

Look at it the other way around: why are all simple, tedious, repetitive jobs not already automated? It's either because it's possible but not economically feasible, or no-one's tried yet... or because it's really difficult and so far there's no solution.

This last category represents a lot of 'low end' jobs. GPTs are unlikely to move the needle here, as appropriate robots to carry out the tasks don't exist yet.

Let's imagine a world where most jobs that cannot be automated are low end, low skill, low pay jobs... what does this world look like economically? In my eyes this is a world where corporations and the extremely wealthy suck up all the economic benefits and we're left with massive wealth disparity. Assets keep going up in price, but there is no means of monetary velocity to the bottom earners.

I don't disagree (notwithstanding the extent to which we're already in a "world where corporations and the extremely wealthy suck up all the economic benefits and we're left with massive wealth disparity").

My point is much more narrow (and simple): that there is a subset of low-end jobs which are currently unautomatable due to their nature, and this will likely not change via the GPTs. This may (tending to will, given sufficient time) change via other technological advancements.

It's meaningless to compare a LLM to a human, anymore than it is to compare a wheelbarrow to a human. To do so betrays both a fantasy projected onto LLMs and a lack of understanding of humans.

We are discussing a system that is very interesting precisely and only because it produces human language that at least superficially looks as though it was produced by a reasonably smart and educated human. For one thing, I don't know how one would disabuse someone of their fantasy projected onto LLMs and a lack of understanding of humans without making comparisons (while, of course, being wary of anthropomorphizing the former.)

A wheelbarrow assisted human can haul 100x faster than a human assisted human.

There, I just compared humans and wheelbarrows in a meaningful way.

There's no problem doing the same with LLMs.

The irony is I think GPT4 gives the best answer. It even gets the idea that language is inextricably tied up with our experiences of reality that most humans seem to simply not understand.

GPT4

"The statement that language models have a "model of reality" can be both true and false, depending on how you interpret "model of reality."

Language models like ChatGPT do have a kind of "model" in a broad sense. They have been trained on large amounts of text data from the internet and have learned statistical patterns from these data. This allows them to generate text that can seem sensible and knowledgeable about various aspects of reality.

However, this "model" is more akin to a very sophisticated pattern recognition system. It doesn't possess a deep or holistic understanding of reality as humans do. It doesn't form beliefs, doesn't have a consciousness, and doesn't experience the world in any sense. It can't learn from the environment or update its understanding over time. Its "knowledge" is fixed at the time it was last trained (for GPT-4, this is as of September 2021), and it can't incorporate new information or discoveries unless it's trained on new data.

Therefore, while language models like GPT-4 can simulate a kind of "knowledge" or "understanding" based on patterns in the data they were trained on, they don't have a model of reality in the sense of an evolving, interactive understanding based on sensory experience and reflection.

So, language models have a "model of reality" only in the sense that they mimic certain aspects of human language use, which is inextricably tied up with our experiences of reality. But this is quite different from possessing a human-like understanding or model of the world."

Haha this scenario is perfect and tells you everything you need to know about the state of AI.

We’re sitting here wondering if the language model also models reality, when we can just ask the thing itself!

We are living in an Asimov novel

GPT's model may be limited to language based, but the vast volume of knowledge it has accessed despite language only, is way beyond any individual human can possibly acquire in a life time. That fact alone makes the GPT world model, or whatever it might be, not to be underestimated. This partially explained why we don't feel GPT's answer is not terribly off mark, even brilliant sometimes.