Hacker News Board Hacker News, browsed like a classic message board
Live data from Hacker News

ChatGPT Plugins

openai.com

Post reply on HN
Print view View original

741–750 of 1001 posts

Re: ChatGPT Plugins

#741

Post by TeMPOraL »

mwigdahl wrote: post #497
kmeisthax wrote: post #321

Earlier quoted context omitted.

AI safety research posits that there are certain goals that will always be wanted by any sufficiently smart AI, even if it doesn't understand them anything close to like a human does. These are called "instrumental goals", because they're prerequisites for a large number of other goals[0]. For example, if your goal is to ensure that there are always paperclips on the boss's desk, that means you need paperclips and so…

There is a whole theoretical justification behind instrumental convergence that you are handwaving over here. The development of instrumental goals depends on the entity in question being an agent, and the putative goal being within the sphere of perception, knowledge, and potential influence of the agent. An LLM is not an agent, so that scotches the issue there.

Agency is overrated. The AI does not have to be an agent. It really just needs to have a degenerate form of 2): a selection process. Any kind of bias creates goals, not the other way around. The only truly goal-free thinking system is a random number generator - everything else has goals, you just don't know what they are.

See also: https://en.wikipedia.org/wiki/The_purpose_of_a_system_is_wha...

See also: evolution - the OG case of a strong optimizer that is not an agent. Arguably, the "goals" of evolution are the null case, the most fundamental ones. And if your environment is human civilization, it's easy to see that money and compute are as fundamental as calories, so even near-random process should be able to fixate on them too.

Re: ChatGPT Plugins

#742

Post by electrondood »

mk_stjames wrote: post #277

I have some odd feelings about this. It took less than a year to go from "of course it isn't hooked up to the internet in any way, silly!" to "ok.... so we hooked up up to the internet..." First is your API calls, then your chatgpt-jailbreak-turns-into-a-bank-DDOS-attack, then your "today it somehow executed several hundred thousand threads of a python script that made perfectly timed trades at 8:31AM on the NYSE whi…

> giving a loaded handgun to a group of chimpanzees.

Hate to be that guy, but this is our entire relationship to AI.

Re: ChatGPT Plugins

#743

Post by bluecrab »

majormajor wrote: post #41
kacperlukawski wrote: post #34

Earlier quoted context omitted.

The fact that the model does not have to rely on its internal knowledge anymore but can communicate literally with any external system makes me feel it may significantly reduce the hallucination.

If it was easy to simply verify truth "with any external system" then would we even need a language model? E.g. if you could just ask [THING] for the true answer, or verify an answer trivially with it... just ask it directly! I ran into this issue with some software documentation just this morning - the answer was helpful but completely wrong in some intermediate steps - but short of a plugin that literally controlle…

All it needs is guardrails which is available already.

Re: ChatGPT Plugins

#744

Post by arvinsim »

pmoriarty wrote: post #728
nwienert wrote: post #442

Earlier quoted context omitted.

We have a ton of protection laws around all sorts of dangerous technology, this is a super naive take. You can't buy tons of weapon technology, nuclear materials, aerosolized compounds, pesticides. These are all highly regulated and illegal pieces of technology for the better . In general the liberal position of progress = good is wrong in many cases, and I'll be thankful to see AI get neutered. If anything treat it…

"We have a ton of protection laws around all sorts of dangerous technology, this is a super naive take. You can't buy tons of weapon technology, nuclear materials, aerosolized compounds, pesticides. These are all highly regulated and illegal pieces of technology for the better." As technology advances, such prohibitions are going to become less and less effective. Tech is constantly getting smaller, cheaper and easie…

> Add in the nearly infinite profit and power motive to get hold of strong AI and it'll almost impossible to stop, as governments, billionaires, and megacorps all over the world will see it as a massive competitive disadvantage not to have one.

I wouldn't say that this is an additional reason.

I would say that this is the primary reason that overrides the reasonable concerns that people have for AI. We are human after all.

Re: ChatGPT Plugins

#745

Post by ShamelessC »

fassssst wrote: post #423
johnfn wrote: post #62

A couple (wow, only 5!) months ago, I wrote up this long screed[1] about how OpenAI had completely missed the generative AI art wave because they hadn't iterated on DALL-E 2 after launch. It also got a lot of upvotes which I was pretty happy about at the time :) Never have I been more wrong. It's clear to me now that they simply didn't even care about the astounding leap forward that was generative AI art and were in…

Then again, the new DALL-E model just released in Bing Chat is really good. Disclosure: I work at Microsoft.

Is it better than Stable Diffusion 1.5?

Disclosure: No job, do whatever I want.

Re: ChatGPT Plugins

#746

Post by dingclancy »

celestialcheese wrote: post #39

This is a big deal for openai. Been working with homegrown toolkits and langchain, the open source version of this, for a number of months and the ability to call out to vectorstores, serpapis, etc, and chaining together generations and data-retrieval really unlocks the power of the LLMs. That being said, I'd never build anything dependent on these plugins. OpenAI and their models rule the day today, but who knows wh…

> That being said, I'd never build anything dependent on these plugins. OpenAI and their models rule the day today, but who knows what will be next.

You cannot assume what will happen in Web 2.0, mobile, iPhone, will happen here. Getting to tech maturity is uncertain and no one understands yet where this will go. Only thing you can do is build and learn.

Whan OpenAI is building along with other generative AI is the real Web 3.0.

This seems to be the start of a chatbot as an OS.

Re: ChatGPT Plugins

#747

Post by hackernewds »

throwaway675309 wrote: post #541
johnfn wrote: post #62

A couple (wow, only 5!) months ago, I wrote up this long screed[1] about how OpenAI had completely missed the generative AI art wave because they hadn't iterated on DALL-E 2 after launch. It also got a lot of upvotes which I was pretty happy about at the time :) Never have I been more wrong. It's clear to me now that they simply didn't even care about the astounding leap forward that was generative AI art and were in…

No that wasn't what they had in mind at all, it was pretty clear from the start that they intended to monetize DALL-E. It's just that it turned out that you require far smaller models to be able to do generative art, so competitors like stability AI were able to release viable alternatives before OpenAI could establish a monopoly. Why do you think that Sam Altman keeps calling for government intervention with regards…

Ironic given OpenAI's initial messaging explicitly was:

"OpenAl is a non-profit artificial intelligernce research company. Our goal is to advance digital intelligence in the way that is most likely to benefit humanity as a whole, unconstrained by a need to generate financial return. Since our research is free from financial obligations, we can better focus on a positive human impact."

Ultimately govt is an idiosyncratic, capricious (and sometimes corrupt?) God of what Nassem Taleb would call Hidden Asymmetries"; as in the case of which elonymous company ingests massive tax credits; or which banks get to survive etc

Re: ChatGPT Plugins

#748

Post by hackernewds »

qup wrote: post #332
chrispogeek wrote: post #297

Earlier quoted context omitted.

Owning up to the "wrong" is good in my book

That's just some weird moral compass. It's almost totally irrelevant if people own up to bring wrong, particularly about predictions. I can't think of a benefit, really. You can learn from mistakes without owning up to them, and I think that's the best use of mistakes.

The opposite of that is sticking to your statements which is stubborn and foolhardy. Owning up to it is courageous.

Which actually lends me to respect politicians who do that, and instead ridicule people who post old videos of Joe Biden or Obama or Hillary Clinton mandating heterosexual couples. A virtuous person is also open to adapting their convictions continually based on present day evidence and arguments - what is science otherwise?

Re: ChatGPT Plugins

#749

Post by hackernewds »

crakenzak wrote: post #384
johnfn wrote: post #62

A couple (wow, only 5!) months ago, I wrote up this long screed[1] about how OpenAI had completely missed the generative AI art wave because they hadn't iterated on DALL-E 2 after launch. It also got a lot of upvotes which I was pretty happy about at the time :) Never have I been more wrong. It's clear to me now that they simply didn't even care about the astounding leap forward that was generative AI art and were in…

Agreed. This makes me realize that OpenAIs leadership is able to look long term and decide where to properly invest, as most of the decisions to take the projects in these directions were made >1 year ago. One can only wonder what they’re working on at this very moment.

Hopefully they do not feel free the pressure to "move fast and break things" because this in turn has the ability to "move fast and break everything"

Re: ChatGPT Plugins

#750

Post by thoop »

dougmwne wrote: post #30

I would be interested to play with a long term memory plugin. It could be a note-taking system that would summarize prior conversations and pull their context into the current conversation through topic searches. This would enable the model to have a blurry long term memory outside of the current context. I played with some prompts and GTP-4 seems to have no problem reading and writing to a simulated long term memory…

I saw this on Twitter that seems to do what you want: https://www.rewind.ai/

I haven’t used it but your comment reminded me of it!

Post reply on HN