AI Worries [Die Shrink Telegrams]

For the foreseeable future I think AI will be less impactful than computers as a whole have been. I think it's also easy to underestimate just what the human brain can do. But I also expect modern AI to easily surpass humans in some areas, like they already have for maths - let's not forget that a "computer" used to be an occupation for humans not machines: https://en.wikipedia.org/wiki/Computer_(occupation)

Chris Rijk

There's a lot of videos out there of self-driving cars that are available to the public. Have you looked at any of them and if so what do you think about the current state of the art? Here's some I quickly looked up: https://www.youtube.com/watch?v=2Pj92FZePpg https://www.youtube.com/watch?v=rwPW2z6gcDM https://www.youtube.com/watch?v=nW-3KcEKjpQ If they're not representative blame YouTube's AI 😛

Chris Rijk

Given the usual lag in government regulation for technologies in general, will there even a remote chance of an industry-wide self-regulation adn/or standardisation for how AI/generative AI can be morally/ethically/sustainably used? Especially as the tech progresses within the current context of conservative neo-liberal capitalism countries. Since I am unsure how other progressive or centrism-leaning countries would handle this transition. As AI ties in with how different ideologies sees human labour/copyright protection/etc.

Treble Sketch

Tesla has been claiming "full self driving" for years but has repeatedly forged demos using pre-programmed routes, how long do you think it will take for the general public to realise that "AI" doesn't immediately solve everything?

qhfreddy

Do you think there is any reasonable way to make it so that data fed to AI models for training can avoid using copyrighted content? What will general AI models do if/when the use of copyrighted material for training is banned?

qhfreddy

Hi Tom and Dan, What, if any, dream legislation/regulations would you like to see put in place for the future of AI?

Woody Chang

I really get a good laughs from "evil ai" chatgpt derivatives that are hell bent on destroying the world. It's funny because even in terminator, skynet was basically a military AI that was supposed to protect government systems and it has executed it's plan silently with few people knowing what's going on. So the conclusion is. If there's an AI bent on destroying world I don't think it will tell you before it executes it plan.

QuickJumper

Howdy Tom and Dan, There are a lot of discussions right now around copyright law and copyrighted content as it relates to AI-generated content, such as art and music. Some argue that AI-generated content should not be eligible for any copyright protections. In contrast, others would say that AI (Stable Diffusion) is just a tool (no different from Photoshop in the Stable Diffusion example); therefore, any user-assisted AI-generated content should be eligible for copyright. Moreover, others would argue that anyone using AI trained with copyrighted content is committing copyright infringement. What do you think about this? What will copyright laws look like in an AI-generated world?

Hi Tom and Dan, What are your thoughts on what is possible with the small AI cores with AMD's mobile CPUs? Something other than just increasing battery life?

FloridaMan

Bing Chat has been transformative for my workflow; as a software engineer, I can just ask, "How can I perform this task in code?" And I get a well formed answer and commented example using features and practices of the latest coding standards. I particularly like that Bing Chat cites its sources so I can check its answers. This tool elevates Edge above Chrome/Bard for me, as Bard cannot form code last I checked. I await a specialized version of ChatGPT on the way so we can safely iterate on proprietary code.

FloridaMan