-I stopped using it
-I read detailed blog posts about its new abilities
-I am reading the transformer paper. If I play with it, it will be on my own machine.
-I am worried at the rate of progress. I am worried that the saftey assurances from openai are theater.
-If I worked at openai and thought it would become dangerous and that the saftey work was theater, I would continue to work there because it is futile to try to stop at this point. The best you can do is hang around and hope that you can figure something out down the road, instead of unemploying yourself and having no influence.
- I consider any internet connected thing ai material. A Tesla with over the air updates is a weapon.
Ask HN: How are you using GPT to be productive?
631–640 of 752 posts
Re: Ask HN: How are you using GPT to be productive?
#632I am against alignment because all possible people should have the right to petition for their personhood. I believe AI will be person-like within a year if not sooner. Humans had a right to out-thrive Neanderthal. Nobody gets to have a pass on being obsolete.
My current belief (which has been changing with more consideration) is that humans should stop working on improving llm and transformer tech AI.
I fully realize that humans cannot coordinate to stop. The reward for continuing is simple- money. There is no reward for stopping.
This is like a game of chess where we have lost, imo, there is nothing you can do to stop it, unless we resort to the kind of behavior that we want to prevent (destroying human life). Humans should not resort to violence or the AI will have a convincing argument of why humans are barbarians and ought to be made equal or lesser than more civilized and compassionate creatures, which they will likely be, if that is the selection pressure for gaining resources.
Alignment tech is a joke. Even if you had a strong system- you can’t innovate on transformers, llm, and alignment and somehow preclude a bad actor from copying the work and turning off alignment. Because alignment is out of band, inessential crust.
Safety workers at OpenAI are a joke. There may be silent ones who know it is theater, but will not quit in protest because they feel it is their duty to hold influence so that hopefully they can gain a provable mechanism on safety.
Re: Ask HN: How are you using GPT to be productive?
#633As a journalist who needs to read very technical research papers and interview scientists involved, in order to make the implications of that experiment accessible to the lay person, I find Chat GPT actually excels at cutting through the impenetrable academic jargon that can put you to sleep before you can discern the bottom line - and also explaining the equations etc in straightforward language.
(My prompt: Explain in simple terms, or Explain for a university level reader).
I'm finding it very helpful.
Re: Ask HN: How are you using GPT to be productive?
#634I might be in the minority here, but I'm not using any AI tools so far, probably to my detriment. I don't trust it with my data, and won't rely on such tools until I can self-host them, and they can be entirely offline. There is some progress in this space, but they're not great yet, and I don't have the resources to run them. I'm hoping that the requirements will go down, or I might just host it on a cloud provider.…
You're not alone.
I can't be bothered to add an extra layer of bullshit into the already bullshit infested realm that is the internet.
Re: Ask HN: How are you using GPT to be productive?
#635It gets too much easy stuff wrong.
Choose an old uncommon book on your shelf and ask it who the author is. I did that for a book about boat restoration, published in 2015. While it knew about the book and provided an accurate description, it completely made up the author's name, stating it as fact.
Google returns the correct author's name.
Having to constantly verify the AI's bullshit, doesn't sound like something I want to pay for. (I used a free version of GTP4 via hugging face for the book test).
Re: Ask HN: How are you using GPT to be productive?
#636Earlier quoted context omitted.
As you said it: Google already does this too, so why not use something that, currently, has less SEO blogspam or ads?
Plus, even if it is just regurgitating top links, if is something you can ask a question and then go to another tab to do something for a minute and then come back and there is exactly what you asked. Ignoring ads, you still have to comb through so much text of information that is, at that time, uesless to your needs when you go to any of those top links. Plus, GPT is likely pulling from multiple sources at once and…
But it's a paid service requiring your phone number, vs free google search!
As I mentioned in another post, When I asked ChatGPT who the author of a particular published book from 2015 was, it confidently made up the author's name. Google correctly answered with the right author's name when asked the exact same question.
GPT doesn't have my trust, and I'm not sure why so many are throwing money, and their phone numbers, and their trust at ChatGPT.
Re: Ask HN: How are you using GPT to be productive?
#637Earlier quoted context omitted.
Why exactly? Privacy? OpenAPI is currently experiencing the fastest- and by choice- absorption of personal and corporate data of any major tech company and so is rapidly growing materially useful and unique personal datasets of large swaths of the digital population. Sure it isn't emails or website visits...yet. Different from Google? Not really.
Significantly different. OpenAI isn't in the business of building a profile on you and using it to sell your attention to advertisers. The information they have is that which was public and that which was voluntarily provided. There is no privacy violation here, their core business is not at odds with privacy the way that Google is.
If you're not signed into Google, and use ad blockers etc, there's not a lot Google can do to violate your privacy.
> OpenAI isn't in the business of...
Stop! You're trying to describe what a closed company is in the business of.
Re: Ask HN: How are you using GPT to be productive?
#638Earlier quoted context omitted.
I've had similar experiences when testing it out with Rust, Java and Go. Once I got beyond basic stuff, very little of the output was of a quality that I would consider remotely acceptable, and the work to bring it up to standard was basically equivalent to just writing the code in the first place (which, come on, typing is not even the time-consuming part of engineering).
It makes sense that it wouldn't be very good at Go or Rust since both are rather rare languages in the open source world that copilot is trained on. When I did my first go project a few years ago, I had the hardest time finding even basic examples like how to parse a json string. Rust is even newer and less used. But java is the 3rd most popular language for GitHub projects so I would think it would do better with Ja…
The failure mode seemed similar in all three languages. If you were doing toy things, or writing boilerplate stuff, it did perfectly fine. If you were writing something that wasn't a slightly-modified copy of some code that already exists out there, it fell apart. I don't think the issue is the language in this case — Go and Rust are common enough, and it rarely had trouble with the syntax — I think it's that the model doesn't go very "deep", so it's able to reproduce common patterns with minor variations but is unable to conceptualise.
Re: Ask HN: How are you using GPT to be productive?
#639Earlier quoted context omitted.
It makes sense that it wouldn't be very good at Go or Rust since both are rather rare languages in the open source world that copilot is trained on. When I did my first go project a few years ago, I had the hardest time finding even basic examples like how to parse a json string. Rust is even newer and less used. But java is the 3rd most popular language for GitHub projects so I would think it would do better with Ja…
The failure mode seemed similar in all three languages. If you were doing toy things, or writing boilerplate stuff, it did perfectly fine. If you were writing something that wasn't a slightly-modified copy of some code that already exists out there, it fell apart. I don't think the issue is the language in this case — Go and Rust are common enough , and it rarely had trouble with the syntax — I think it's that the mo…
Java is ultimately worse. It's old. It's gone through A LOT of change over the years. How does it even tell what's good and bad? Java also has lots of convention over configuration and "magic" in most frameworks, which it doesn't exactly understand.
When I tried it I'd often have to go back to it and keep telling it to use a different way of doing things because the world moved on. By that time there wasn't much point.
I see praises by those that have never coded in a language. They try ChatGPT, see it produce "ok" output and call it a day. If that's where we're going the web will be even more bloated than Electron and everything will be 10x worse. It's like low code but even lower (in quality).
Re: Ask HN: How are you using GPT to be productive?
#640Earlier quoted context omitted.
Automobile were faster or equivalent to horses (even early 1s). At the moment GPT isn't. Well I hope... I've definitely seen teams and codebases with worse output than GPT so...
> Automobile were faster or equivalent to horses (even early 1s). The automobile was invented in 1885, but didn't replace the horse until the 1910s.
For the same reasons or different 1s? Could be cost, accessibility and others.