The A.I. Thread

[GGG]

Originally Posted by Itse

For me, the problem isn't tools. The problem is that AI's (as they are) are most useful when asked to produce large quantities of something of unknown quality, fast, or when asked to create tons of new art, or when imitating human behavior.

These aren't really things we're short of.

We also aren't set up in a way where people working more effectively will have to work less or get to earn more. People are more productive than ever, but people are still working themselves to death, and only the rich are getting richer.

That’s just a taxation problem. If AI can make 50% of jobs redundant then that’s awesome. Just have to tax the surplus that is created.

[Itse]

Originally Posted by GGG

That’s just a taxation problem. If AI can make 50% of jobs redundant then that’s awesome. Just have to tax the surplus that is created.

Edited my message somewhat after you quoted, sorry :P

I agree, theoretically, but we've already been at that point for decades, and it was much easier to make political changes when labor had value.

AI's will cause the value of a lot of labor to just plummet, and the way our society is currently set up, that means most of us will be of constantly decreasing interest to people with power and money.

Oh, and AI soldiers are just around the corner, so good luck with that revolution.

[JohnnyB]

https://twitter.com/user/status/1650935144100798475

[Bill Bumface]

Originally Posted by Fuzz

https://www.theregister.com/2023/04/...insecure_code/

This was on 3.5, but serves as a good warning. Hopefully we don't have too many people making insecure apps handling private data and payment information.

Humans write insecure code all the time. If your security stance relies on humans to get it right, you're doomed anyway.

Anyone pushing commits from ChatGPT should be subject to the same automated security checks they were as flawed humans.

[TorqueDog]

Yep -- using ChatGPT's code moves your role from being the code writer to the code reviewer. You still have to do a good job of validating that the code is secure and fit for the purpose you're intending to use it for.

[Fuzz]

The news-rating group NewsGuard has found dozens of news websites generated by AI chatbots proliferating online, according to a report published Monday, raising questions about how the technology may supercharge established fraud techniques.



The 49 websites, which were independently reviewed by Bloomberg, run the gamut. Some are dressed up as breaking news sites with generic-sounding names like News Live 79 and Daily Business Post, while others share lifestyle tips, celebrity news or publish sponsored content. But none disclose they’re populated using AI chatbots such as OpenAI Inc.’s ChatGPT and potentially Alphabet Inc.’s Google Bard, which can generate detailed text based on simple user prompts. Many of the websites began publishing this year as the AI tools began to be widely used by the public.

Giansiracusa, the Bentley professor, said it was worrying how cheap the scheme has become, with no human cost to the perpetrators of the fraud. “Before, it was a low-paid scheme. But at least it wasn’t free,” he said. “It’s free to buy a lottery ticket for that game now.”

https://www.bloomberg.com/news/artic...-content-farms


RIP internet. We hardly knew thee.

[dammage79]

I'm surprised this thread hasn't been bumped with the news of Hinton quitting and pleading for everyone to slow down this AI train.

[JohnnyB]

Who is going to slow it down? Any tech that has utility does not go back in the box. Add on to that the speed of change and improvement in AI, the potential benefits of being the market leader, and the AI arms race between the world's two biggest powers and there's a lot momentum carrying it forward.

I really respect Hinton and his views, but regulators can't keep up with things moving at half this speed. I also tend to think people have a lot less control over these kinds of things than we would like to think we do.

[Fuzz]

Ya, we bought the ticket, no getting off the ride now. But I do appreciate actual experts expressing serious concern. At least we can't say "nobody warned us!"

[Itse]

You can also just move the development to a country without regulation.

Or not even the development, just the servers.

[Wastedyouth]

The singularity is nigh.

[Firebot]

You can be sure that if the US / Canada / Europe banned AI, China would further develop it, and in a world where we promote globalism and outsourcing it will quickly be a losing battle. And so many major corporations onboard with it, it will only accelerate.

On a side note, I've officially recently hit the 100$ generation milestone directly generated from AI (my app currently on hold for this other proof of concept idea I had to get it going as it was too good, but going back for the finishing touches on the app), it's a highly scalable website based business and subscription based only a week in meaning steady passive income and succesful proof of concept, (app will be subscription based as well). I will update once I hit the 1K, 10K and 100K mark over all my projects (multiple streams). The 1K mark will likely be hit with this alone within 3 months based on current growth.

Originally Posted by Firebot

And I am this far in with zero app programming experience with nothing installed on my PC, from an idea GPT4 came up with. This is also outside of 2 separate passive revenue streams proof of concepts I already started with it but still need to prove. I will update once I get a major breakthrough or milestone with it.

Not even a month in since I posted that, not bad.

[Russic]

Really interesting article purported to be an internal Google document about how AI is poised to become an unrestrainable force that could make it hard for a big company to control and wield. The idea of training something on a smaller amount of data for cheap is super appealing.

https://www.semianalysis.com/p/googl...at-and-neither

[Fuzz]

Originally Posted by Firebot

You can be sure that if the US / Canada / Europe banned AI, China would further develop it, and in a world where we promote globalism and outsourcing it will quickly be a losing battle. And so many major corporations onboard with it, it will only accelerate.

On a side note, I've officially recently hit the 100$ generation milestone directly generated from AI (my app currently on hold for this other proof of concept idea I had to get it going as it was too good, but going back for the finishing touches on the app), it's a highly scalable website based business and subscription based only a week in meaning steady passive income and succesful proof of concept, (app will be subscription based as well). I will update once I hit the 1K, 10K and 100K mark over all my projects (multiple streams). The 1K mark will likely be hit with this alone within 3 months based on current growth.



Not even a month in since I posted that, not bad.

What does it do? Can you post it?

[Kipper_3434]

People who start threads with paid links, should be permanently banned.

[jayswin]

Originally Posted by JohnnyB

Who is going to slow it down? Any tech that has utility does not go back in the box. Add on to that the speed of change and improvement in AI, the potential benefits of being the market leader, and the AI arms race between the world's two biggest powers and there's a lot momentum carrying it forward.

I really respect Hinton and his views, but regulators can't keep up with things moving at half this speed. I also tend to think people have a lot less control over these kinds of things than we would like to think we do.

Originally Posted by Fuzz

Ya, we bought the ticket, no getting off the ride now. But I do appreciate actual experts expressing serious concern. At least we can't say "nobody warned us!"

People expressed concern decades ago. That was the time to act. Now we're headed for exactly what we've been warned we'd head towards.

[Firebot]

Bumping this as Open AI announced earlier this week it's rolling out browsing and plugin capabilities to all chatgpt plus users. I got browsing capability a few days ago, and just yesterday found out I had plugin access as well.

Interestingly enough as of right now, you cannot combine both as of yet.

Browsing...is definitely beta. The biggest issue seems to be that it is unable to scan many pages due to pages blocking bots. It also can get sidetracked, and since it doesn't know what is legit or not as a source and goes based off the first google result (or bing), the result could be worse than if it did not browse at all. It also cannot read PDFs via browsing. It can however provide an extremely detailed and accurate summary of the Ukraine war and have access to real time info which is a great plus.

Plugins already are looking like an App Store. You have OpenTable, Kayak, Expedia etc. I used Zillow which gave me house comparable in an area I am looking at, which definitely was interesting. Not sure that I would use through chatGPT though.

By far the most useful right now would be Wolfram and chatwithPDF (which actually allows browsing with a direct URL to a specific PDF page). This allows specific new use concepts that weren't possible with the base chatgpt including one monetization concept I had. Plugins are rolling out extremely quick, and gpt 4 with plugins seems to be the way forward.

Wolfram specifically fixes a lot of gpt 4 inherent shortcomings, most notably math without the ability to validate itself. Now it has the ability to get accurate results including real time data if wished.

Deeplearning.ai has a free prompt engineering course in collaboration with OpenAI. This course uses the API functionality via python (Still waiting for my GPT4 API key but can do this with gpt3.5) but most practices also applies to chatgpt. Note that API calls costs $$$ if you do run them.

https://www.deeplearning.ai/short-co...or-developers/

Anyways, good time to sign up to chatgptplus if you were waiting on better features.

[chemgear]

https://gizmodo.com/zoom-ai-privacy-...ata-1850712655

Zoom Contradicts Its Own Policy About Training AI on Your Data

Zoom updated its Terms of Service in March, spelling out that the company reserves the right to train AI on user data with no mention of a way to opt out. On Monday, the company said in a blog post that there’s no need to worry about that. Zoom execs swear the company won’t actually train its AI on your video calls even though the Terms of Service still say it can.

[FlamesAddiction]

Microsoft's AI-written travel guide called Ottawa Food Bank a 'beautiful tourist attraction'

https://twitter.com/user/status/1692233111260582161

“In Ottawa, you will find some beautiful attractions that you just cannot miss! Places like The Winterlude Festival, National War Memorial, and Ottawa Food Bank and many more,” the article begins.

The food bank was listed as the top three attractions.

“We observe how hunger impacts men, women, and children on a daily basis, and how it may be a barrier to achievement,” the article stated below an image of the Ottawa Food Bank.

“People who come to us have jobs and families to support, as well as expenses to pay. Life is already difficult enough. Consider going into it on an empty stomach.”

https://nationalpost.com/news/ottawa...k-ai-microsoft

[Fuzz]

It was reported that the company had laid off dozens of journalists and editorial workers in 2020, replacing them with artificial intelligence.

The ####ification of the planet continues.