461–470 of 496 posts
Earlier quoted context omitted.
John Henry beat the steam powered drill after all!
Yes, he did. But working machines vs thinking machines is still a different ball game. I won't give details of why I believe the way I do (I might someday; I have a blog post draft), but I'm extremely confident that thinking machines will never surpass humans.
Earlier quoted context omitted.
I think you're describing the difference of AI creating code vs developing _with_ AI which firmly keeps humans in charge but enabling them to be many times more efficient.
They'll only be more efficient if they don't review the code. Otherwise, these tools are just typing helpers, and typing is a small part of the job.
Earlier quoted context omitted.
GPT4 is certainly not fulfilling the definition of reasoning. It's borrowing the intelligence of every human who wrote something that went into its model. To demonstrate this, ask it to prove something that most or all people believe. Say some "intuitive" math thing. Perhaps the fact that factorial grows faster than exponential functions. And no, don't just have it explain it, have it prove it, as in a full mathemati…
In order for an AI to reason it doesn’t mean it has to be able to reason about everything at any level - most humans cant rediscover fundamental mathematical theorems from basic axioms, particularly if you keep removing them until they fail, but I don’t think that means most humans are unable to reason. Take this problem instead which certainly requires some reasoning to answer: “Consider a theoretical world where pe…
Because Ben is the tallest, his feet are the biggest, and because he takes the same amount of steps as the others, the amount of area he steps on is larger than the area that the others step on.
Therefore Ben is most likely to be the one to step on the most bugs.
Easy. And I'm not brilliant.
The problem with testing these tools is that you need to ask it a question that is not in their training sets. Most things have been proven, so if a proof is in its training set, the LLM just regurgitates it.
But I also disagree: if the "AI" can't reason about that, it can't reason because that one is so simple my pre-Kindergarten nieces and nephews can do it.
But even if not, the LLM's should have "knowledge" about exponential functions and factorial because the humans who wrote the material in their training sets did. So it's not a lack of knowledge.
And I claim that most humans could rediscover theorems from basic axioms; you've just never asked them to.
Earlier quoted context omitted.
I’ve been using GitHub copilot and ChatGPT at work for a year now. I think you’re wrong. You don’t just wily nilly accept code from the AI anymore than you would from stackoverflow. But it would also be silly to ban yourself from copying code from stackoverflow. Don’t trust any code, regardless of the source, and don’t stop using your brain.
With how bad SO has gotten, I actually don't copy code from there anymore.
Earlier quoted context omitted.
Yes, he did. But working machines vs thinking machines is still a different ball game. I won't give details of why I believe the way I do (I might someday; I have a blog post draft), but I'm extremely confident that thinking machines will never surpass humans.
Does it relate to your religious beliefs?
I don't fully understand the why yet, but I do understand a lot of it, enough to be completely confident in this.
I'm working on the rest, and once I do understand it, in a scientific sense, I'll write and release the blog post.
Earlier quoted context omitted.
>1) That is a horribly denigrative view of human creativity, one that simply does not reflect reality. No it's not. It's a realistic view. Do not discount a view just because it's not inline with positivity. Only the realism of the view should be considered. And unfortunately it does reflect reality. There's a huge amount of data supporting my viewpoint. Most of the content right now on the internet in books and in m…
> There's a huge amount of data supporting my viewpoint. Most of the content right now on the internet in books and in movies are generic human generated content. You literally picked the closest thing to AI-generated content, soulless statistics-driven garbage, literally the first thing that will be automated by AI in corporate media ;) > You can be creative all you want, it doesn't mean other people will care or wa…
Earlier quoted context omitted.
It's a difference of degree. I still check the output of the spell checker. I still check the output of the search engine. The problem with these "AI" tools is that they are magnitudes more likely to encourage you to not check the output because checking the output takes orders of magnitude longer than it does to check the output of a spell checker or search engine. Do people refer to spell checkers as AI? No, so I d…
It sounds like your big objection here is to using ChatGPT to generate entire paragraphs of text. I agree: I think it's rude to expect people to spend more time reading something than you spent writing it! But... there are still plenty of ways to integrate ChatGPT-like-tools into a writing process that I think avoid disrespecting your audience like that. I use it as a thesaurus, or to suggest ways my writing could be…
My objection to ChatGPT is yes, about generating full sentences.
I object to it for many reasons, but not necessarily just because it's rude. My biggest concern is actually that it takes out the voice of the human.
I have a hard enough time figuring out what people are thinking in person (I'm on the spectrum) that doing it through text is like trying to sense the movement of fish with my eyes closed on shore. Making everybody have the same bland voice would turn that up to 11.
I want the human, even if the human isn't perfect.
Earlier quoted context omitted.
They'll only be more efficient if they don't review the code. Otherwise, these tools are just typing helpers, and typing is a small part of the job.
Not sure I agree, but to be fair my perspective is informed by this talk I watched not too long ago https://www.youtube.com/watch?v=qmJ4xLC1ObU
First, he claims that Copilot/GPT only needs more data and more compute to get better. I disagree. It needs both, for sure, but I think it needs more.
Also, it won't get any more data! As LLM's are used more and more, the data fed in will be more and more like what they would generate anyway, which will lead to the models overfitting on things and getting worse.
He claims that bots make mistakes quickly and that allows iteration. This is true, but the iteration will probably be more like bogosort than anything intentional. (Bogosort is famously very expensive.)
Why is it like Bogosort? Because even good "prompt engineers" are more or less creating enchantments on the fly to coax information out of a black box. To me, that seems like a more or less random search. Hence, random search, random sort, it's like Bogosort.
He claims reviewing the code is 100x faster than writing it in his experience. Well, yes, because you don't dive into as much as the original writer did. Reviewers only have so much time, so they spend only as much time as they have. They could spend more and catch more bugs.
He claims that the AI will (eventually) take instructions from you and run them directly. He says they won't generate the code, they are the code.
They can only do this if they are Turing-complete, which if they are the typical neural nets I've seen, they cannot because data can only flow one way. Turing-completeness requires data to be able to flow conditionally, forwards, and backwards, and any combination. These models cannot do that. They would also need recursion.
He claims they have reasoning capabilities. I claim they only have the appearance of reasoning, borrowed from the reasoning capabilities of the humans that wrote the material used in the training sets.
His example of the cards is good, but not that impressive. It didn't tickle any Turing-completeness.
Those are my thoughts as I watched it. It was pretty good though. It was convincing. I know I'm weird in that I just cannot be convinced.
.
But we're missing an important point, as far as software development or at least web development/user-centered apps goes at least, if not many other situations; We will need less software, not more, and the software we will need will be much more infrastructure and 'industrial'.
For example, currently I might go to a units conversion website to do a quick cooking calculation. chatGPT.
I go to a weather website (always via google, because I never remember the website names) to check the weather. chatGPT.
I want some information on my stocks, and their outlook. chatGPT - can summarize the price movements and the 'buzz' around said stocks without wading through pages of web UI.
I want to book a holiday. Instead of wading through ad- and promotion- laden booking sites, chatGPT.
The authors example of 'writing articles'. Why bother reading them? Unless it's reading for pleasure, do I want to seek out a particular publication and wade through a load of fluff to get the meat of what I want to inform myself about? No, just ask chatGPT.
Sure it's going to have to learn to present results neatly in tables and pictures and so on, and it's going to have to use plugins to get actual answers to some things. And I don't want to have to type at it all the time, so I'll want some 'promptmarks' or something more sophisticated.
But all these websites that we're worried GPT will write for us, unemploying developers - it's 'worse' than that - the sites won't be needed to start with.
As for other stuff, say news, why bother getting chatGPT to write a styory for a page on NYT or whatever. I can just ask GPT 'whats up' and it'll tell me news.
Sure for all this, it's not quite there yet, but I think the writing is on the chat-box, er, wall.
.
The other main issue being missed is capitalism. We've enjoyed a golden age of creation where basically all you need is a laptop and some open-source software, and you can code, or write articles, draw artwoks or whatever. Albeit you have to use infratructure to distribute that code or make it available online etc.
With AI, the means of production will suddenly change from a laptop, to multi-million dollar models, owned by someone else. That could have a profound effect on who gets to create, and under what conditions.
Earlier quoted context omitted.
AI driven business will be cheaper, and faster. The "better" factor will be a case by case situation. One could argue that your competitor combined with their skills, expertise, AND AI would have a market advantage over you.
And my bet is that they will not, long term. If you use "AI" to increase productivity, you're doing it by reviewing the code from the "AI" less than you would your own. That lack of review, and testing, will slow down velocity over time as the software becomes untenable, unmanageable, and too complicated to understand. Meanwhile, my slow velocity at first will result in great design and little technical debt. Eventua…