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

GPT-4

openai.com

Post reply on HN
Print view View original

671–680 of 1001 posts

Re: GPT-4

#671

Post by make3 »

d3ckard wrote: post #600
hooande wrote: post #584

After watching the demos I'm convinced that the new context length will have the biggest impact. The ability to dump 32k tokens into a prompt (25,000 words) seems like it will drastically expand the reasoning capability and number of use cases. A doctor can put an entire patient's medical history in the prompt, a lawyer an entire case history, etc. As a professional...why not do this? There's a non-zero chance that i…

I must have missed the part when it started doing anything algorithmically. I thought it’s applied statistics, with all the consequences of that. Still a great achievement and super useful tool, but AGI claims really seem exaggerated.

it definitely learns algorithms

Re: GPT-4

#672

Post by trts »

OkGoDoIt wrote: post #533

From the livestream video, the tax part was incredibly impressive. After ingesting the entire tax code and a specific set of facts for a family and then calculating their taxes for them, it then was able to turn that all into a rhyming poem. Mind blown. Here it is in its entirety: --- In the year of twenty-eighteen, Alice and Bob, a married team, Their income combined reached new heights, As they worked hard day and…

If automation can make tax code easier to be in compliance with, does this imply a reduced cost of increasing complexity and special exceptions in the tax code?

Re: GPT-4

#673

Post by make3 »

drumhead wrote: post #663

Are they going to limit access to this because they think its too "dangerous". That would be a tragedy if they did. We've seen how opening access up to as many people as possible has produced some of the best results and demonstrated the usefullness of these LLMs. They need to get it out to the public as soon as possible and then see what the public come up with. I really feel like a new age of innovation is upon us…

they haven't given any sign that they will limit the access. They have given signs that they are capitalists & are ready to do a lot to make money, like not putting a list of authors on the GPT4 paper & not write anything about the model architecture or training process

Re: GPT-4

#674

Post by varenc »

antondd wrote: post #660

I'm equally excited and terrified. Excited for the possibilities of a new technological revolution, but terrified for all potential abuses of technology the said revolution would bring. What is stoping our adversaries from developing malicious AI models and unleashing them on us?

> What is stopping our adversaries from developing malicious AI models and unleashing them on us?

That fear is a big part of OpenAI’s reasoning behind not open sourcing their models. So in the immediate terms I’d say malicious uses are limited by its locked down nature. Of course, that’ll eventually end. The key research that makes this possible is open and eventually access will be democratized.

My personal take, which I know is controversial, is that by locking down these models, but still making them available over a GUI/API, the world can better prepare itself for the eventual AI onslaught. Just raising awareness that the tech has reached this level is helpful. Still not sure how we’ll deal with it when the bad actors come though.

Re: GPT-4

#675

Post by throwaway4aday »

Sol- wrote: post #313
nealabq wrote: post #227

Test taking will change. In the future I could see the student engaging in a conversation with an AI and the AI producing an evaluation. This conversation may be focused on a single subject, or more likely range over many fields and ideas. And may stretch out over months. Eventually teaching and scoring could also be integrated as the AI becomes a life-long tutor. Even in a future where human testing/learning is no l…

With the current progress, human learning seems to be obsolete soon, so there's little point in optimizing an AI for teaching. Unless you mean only as a hobby to pass the time. > AIs may be tutoring and raising other baby AIs, preparing them to join the community. Probably I'm not futurist enough, but I'm always amazed at how chill everyone is with supplanting humanity with AIs. Because there doesn't seem to be a pla…

Nah, this is the second part of the industrial revolution. First part replaced and augmented physical abilities so instead of making things by hand we automated away a large portion of the work but not all of it. This is augmentation and automation for intelligence. Yes, a lot of what we currently do "by mind" will be automated but these systems have their limitations. It's still going to be crazy though, imagine what it was like to be the town blacksmith when they first heard of a steam hammer. Nowadays we have very few blacksmiths but we have a lot of people designing parts that will be made on a CNC. What is the role of the human once the labour of clicking away at a mouse hunched over a screen to produce a part is automated? Now we just discuss the end product with the AI, look through some renderings, ask for different versions, ask it to run simulations, tell it to send the file to the CNC? Now that anyone can "design" a part or a whole product by talking to an AI what kind of new jobs does that entail? There might be a big demand for computer controlled production of one off designs. What kind of incredible inventions and wonders can we create now that we can basically conjure our thoughts into existence? There's going to be a whole cross-disciplinary science of combining various areas of human knowledge into new things. Too bad Disney already coined Imagineer.

Re: GPT-4

#676

Post by nbzso »

I don't understand how in the near future this will not remove designers, developers, and especially lawyers and marketers from the workforce. Help me out to conceptualize the future use cases. How about the more "impactful" implementation in creating a version of social index in which the "A.I." will be the Agency?

Re: GPT-4

#677

Post by make3 »

_boffin_ wrote: post #70

This technology has been a true blessing to me. I have always wished to have a personal PhD in a particular subject whom I could ask endless questions until I grasped the topic. Thanks to recent advancements, I feel like I have my very own personal PhDs in multiple subjects, whom I can bombard with questions all day long. Although I acknowledge that the technology may occasionally produce inaccurate information, the…

it makes shit up still

Re: GPT-4

#678

Post by TeeMassive »

amelius wrote: post #595
hooande wrote: post #584

After watching the demos I'm convinced that the new context length will have the biggest impact. The ability to dump 32k tokens into a prompt (25,000 words) seems like it will drastically expand the reasoning capability and number of use cases. A doctor can put an entire patient's medical history in the prompt, a lawyer an entire case history, etc. As a professional...why not do this? There's a non-zero chance that i…

> As a professional...why not do this? Because your clients do not allow you to share their data with third parties?

You mean like the cloud?

Re: GPT-4

#679

Post by lionkor »

amai wrote: post #513

I would love if GPT-4 would be connected to github and starts to solve all open bugs there. Could this be the future: Pull requests from GPT-4 automatically solving real issues/problems in your code?

It still makes up stuff.

Re: GPT-4

#680

Post by whazor »

mk_stjames wrote: post #577

I just finished reading the 'paper' and I'm astonished that they aren't even publishing the # of parameters or even a vague outline of the architecture changes. It feels like such a slap in the face to all the academic AI researchers that their work is built off over the years, to just say 'yeah we're not telling you how any of this is possible because reasons'. Not even the damned parameter count. Christ.

I think it is important to know, as a user, how things roughly work. Now we don't know how they fixed previous flaws or what the drawbacks are.
Post reply on HN