Tech leaders urge pause in 'out-of-control' AI race

At the minimum just talking about a pause on AI is raising awareness of the issue. Throwing up your hands and just allowing AI develop to continue unabated without any consideration of the potential negative consequences is what has gotten humanity into a lot of problems.

 

This is the problem that because so much is done with so little thought of long term consequences.

It’s less the specific technology but this attitude thst will wipe us out. Whether it’s AI, genetic engineering or self replicating nanotechnology.

 

We need regulation on the international level to deal with the development and use of AI. Given both the potential
Promise and threat of AI we need to treat this like nuclear proliferation.

 

We need AI law enforcement of our cities and borders
AI drones weaponized and having facial recognition software

 

Unfortunately we're going to have to see something really bad happen first.

 

and then, of course, it's too late

 

Maybe.

I think something very consequential can happen without it being "too late". Maybe some deep fake disrupts a really important election or something.

We haven't flirted with the whole self-imposed doomsday problem for most of our history, so the book is still out on how we can handle such responsibility.

One thing I do know is that AI is effectively a key to power, and what we know about keys to power is that their desire is universal and evergreen. Nukes are the perfect analogy. We can sit around and talk about 'no more nukes' all we want, but the truth is every world leader knows that a nuclear weapon is an automatic ticket to a seat at the table with the big kids... so they will always be sought.

 

But yet we have, and I'm not talking about Dr Strangelove stuff either, I'm talking Able Archer, Vasily Arkhipov, etc...

https://en.wikipedia.org/wiki/List_of_nuclear_close_calls

https://www.theatlantic.com/interna...closer-to-nuclear-war-than-we-thought/276290/

 

I got caught up reading the links in the top quarter, but the rest is mostly sensible in terms of the immediate impact of AI in general.

So I won't post that part...

spectrum.ieee.org /gpt-4-calm-down
Just Calm Down About GPT-4 Already

 

I'm talking about the length of time human civilization has been around vs. how long we've had access to the "end species" button.

I'm aware we've had close calls. What I'm saying is that we don't have a lot of data points and the nuclear age is a percent of a fraction of a percent in terms of human history.

 

I'm becoming frustrated with the way people express their fears of AI. AI will never be the downfall of society because it became sentient and decided to take over the human race. No, AI is dangerous because it is simply a tool that wealthy people can use to **** over poor people like any other tool.

Already seeing this with pricing algorithms forming cartels and social media algorithms creating mentally unstable individuals because rich people found out that negative discourse is the best way to maximize engagement and therefore profits.

 

Gotcha, we're talking about 2 different things

 

Yea I forget which Sean Carroll podcast where he had an AI expert on that mentioned AI won't really be able to do much without a body to make approximate observations on newly presented data. Thought process is, say you want to mine the dark side of the moon for Helium-3. You do your satellite flybys, collect as much data as you can for your mission as can be had, run it through your AI to calculate best plans, do all the prep work you believe is necessary for the mission to be a success and upon landing on the spot thought best to land on the moon, you realize some new complication upon landing that you couldn't foresee until it was too late, and how a lot of progress on things is a constant updating of new information presented based on the real world that you didn't anticipate.

So you can code AI to anticipate what to do in a scenario a real person or team of people did themselves and taught the AI how to troubleshoot, but when it comes to something completely new that it wasn't trained on that could only be learned by a physical presence there to detect the new thing to troubleshoot, it won't know what to do. However if it did have a body that was extremely sensitive to collecting data about the real world around it, it would be more autonomous and could respond on its own to its real world environment making it more in the mold of what living organisms are more or less pre-programmed to do in our own DNA. DNA that creates human cells with various individual functions that collectively make up our bodies, where for example, we have hands to manipulate the world with nerve endings that connects to our CNS that is connected to our brains which is already pre-programmed to monitor hunger, sleep, and environmental changes like smoke coming from burnt popcorn in the kitchen relayed from our senses to our brain; We are constantly being bombarded and updated with new environmental information our body communicates to our brain that then manages based around that.

AI can't do that. It can't do that without a body with a CNS capable of describing the environment around it. It still needs input from people's decades of scientific research conducted in lab settings by people using their own physical bodies to walk around and conduct experiments that they then physically uploaded onto the internet that was then gathered for AI to be able to spout off for all to be impressed by.

I think the more appropriate fear over AI is the quality of the data it receives and the real world context of that data that it by default is incapable of truly understanding, again because it's absent a body. As the Last Week Tonight on AI touched upon, if your data for example, has a racial bias to it, and you input that data to your AI, then your AI will also have a racial bias to it. That then makes it easier for bigots to wash their hands of something they'd benefit from. It's not them that's being bigoted. It's the AI. Think how landlords rationalize having algorithms based on rent increases being used an excuse to justify their actions.

 

https://www.goodmorningamerica.com/...medical-procedure-paralyzed-man-walk-99588314

Was gonna post this yesterday. A.I. used to help paralyzed man walk

 

This is amazing !

 

We have a big problem with nuclear proliferation but it’s telling that other twice we haven’t used nukes. We haven’t even seen a dirty bomb used by non state actors even though there are a lot of groups that have tried.

 

Yes very much the danger of AI is how people will use it. That alone should give pause for ethics and regulation to develop around.

The possibility of a sentient AI manipulating us while far fetched can’t be ruled out though and given we have so little
Understanding of how AI works that is more reason for ethics and regulation.