I watched a presentation over YouTube yesterday that got me pretty freaked about the potential issues that may arise with the creation of AI.
You can check it out here:
A data point that they continuously return to is:
50% of AI researchers believe there's a 10% or greater chance that humans go extinct from our inability to control AI
They site the 2022 Expert Survey on Progress in AI as the source for this info, specifically answers to the question:
What probability do you put on human inability to control future advanced AI systems causing human extinction or similarly permanent and severe disempowerment of the human species?
When AI experts feel that there is a 1/10 chance of us going extinct from AI, it's cause for concern in my book.
I highly recommend checking out the video. They do a great job explaining how the field of AI has changed over the past 5 years and how/why it's progress is rapidly accelerating. They give specific examples of AI being used nefariously today and talk about the potential harm that may come. They also talk about how it can be so difficult to foresee these issues, comparing this point time to the advent of social media — how we didn't realize that this tech would bring about new problems or amplify existing ones, things like: information overload, addiction, doomscrolling, influencer culture, sexualization of kids, Qanon, and more. These things weren't the intended outcome, but they were very real effects that we're still contending with today.
I really hope that we take more time thinking about the negative effects that AI may cause before deploying these things into the world. We know that this is highly impactful tech with a lot of potential for good, but if we move too fast and don't think about the potential consequences, we may end up with a whole slew of AI-driven issues that we're not prepared to contend with.
Top comments (15)
This is genuinely terrifying. We’ve been so naive with the internet and technology and it’s hard enough to tell your grandparents to stop clicking random links or answering unknown callers. From how quick it’s advancing already! Imagine this time next year.
Oh totally! My mind went to the same place when they talked about how scammers might us AI voice technology to copy a young person's voice, then call that person's parents and ask for their social security number. People are going to learn to trust no one... that doesn't sound like a very nice world to live in.
AI/ML is going to amplify all the problem you see in the world today: it's already been involved in raising rent prices aggressively. And considering how often it perpetuates racism, I can't wait until it's a central part of surveillance technology, criminal justice, and policing.
Ughhh... the raising of rent prices sounds about right. So much of what AI is solving for right now is profit-driven. I really hope that we find more ways to use it to help everyone, not just the greedy rich and powerful.
And you're totally right that we've seen AI perpetuating racism and discrimination. I found a post on the ACLU site that discusses this topic; it's understandably pretty American-centric (it is the American Civil Liberties Union after all) and nearly 2 years old (AI has made a lot of "progress" since then) but it does a decent job highlighting some of the issues you mention, including talking about housing discrimination driven by AI.
With AI, I believe there is massive potential for good, but there is equal potential for abuse. It's certainly complicated. When thinking of AI I have moments of optimism and pessism, but currently am leaning a bit toward the pessimistic side. I really hope we start to act more cautiously.
Sound very pessimistic, but I can't state with any certainty at all that you're wrong about this ...
Right, I know how it sounds, but I think that we have to be ready for any negative effects caused by this. This is essentially automation, and we all know automation can help break things worse than doing things manually
I think you're not wrong - and I was very much intrigued by what you said:
"it's already been involved in raising rent prices aggressively"
I'm not doubting it, you have a source for that? I'd like to know more as I think that might be a case that's VERY instructive (not just for me but for all of us) as to how this whole AI thing might pan out, and the multitude of (largely, or completely) unexpected "side effects" it's going to have in our societies ...
I you could share that, I think it could help us all be forewarned and educated as to the widespread effects this may (make that "will") have.
I see being "educated" as the #1 weapon here, I expect the naive and the gullible tol be hit incredibly hard by this, so let's make sure there are as few as possible of those - we all need to be EDUCATED about this.
P.S. well I did the obvious, a simple google search was my friend again:
reddit.com/r/Futurology/comments/z...
And I read that, and as commenters there rightly said - these AI algorithms are still TOOLS used by humans, with good or bad intentions - as of now, AI does "what we tell it to" ... they're tools, and as any tool it can be used for good or for bad - ultimately, humans are doing this.
Until "the robots" take over and go rampant all by themselves - but I wonder if they'll be as evil as some humans are, lol :)
Story for the delay, I had a lot going on.
A good article on the rent issue:
propublica.org/article/yieldstar-r...
I think we'll have a good idea of the true scale of any issues by the end of the year at the soonest. The number of AI products just announced the last month or so has been more than anyone expected.
I think once any sort of income stream is seen from those AI products/services, there will be increasing efforts and competition for that income, and that's where problems are going to manifest. I don't think we are there yet, but once we hit the tipping point, things will take off, and any problems are going to be shockingly apparent. But once the cat is out of the bag, it'll be nearly impossible to change unless it personally affects the people with the power to change anything.
I personally don't think the danger AI poses will be because of sentience or singularity(although there's been multiple people warning about this, like Janelle Shane - youtube.com/watch?v=OhCzX0iLnOc, Sam Harris - youtube.com/watch?v=8nt3edWLgIg, Jay Tuck - https://www.youtube.com/watch?v=BrNs0M77Pd4&t=1025), but meerly because it's incredibly powerful.
There is a brutal marketing campaign to encourage investors to put money into AI, which consists of saying that this technology is so powerful and influential that it can even put humanity at risk...etc.
The funny thing is that people swallow the advertising campaign like popcorn in a movie theater... Not long ago it was the META, now it's the AI, tomorrow something else...
There is a solution to everything but unfortunately society will be relying on the uncoordinated, ill-informed and possibly corrupted efforts of our elected representatives to implement safeguards. Business will not help anyone.
There will be a huge amount of "fake" unleashed upon us via the internet, that seems already 99% to 100% certain, and that's probably the most innocent effect that AI will have ...
So, if anything, we need education and lots of it, to make people aware and to "weaponize" them against this - it means you simply CANNOT trust anything, I mean nothing, you see on the internet ... maybe that was already the case, but it'll be amplified 10-fold, 100-fold.
If that would be the worst outcome, well, then we're blessed.
I see this as the 5th or 6th "wave", after the Agricultural Revolution, the Industrial Revolution, the Computer/IT Revolution, the Internet Revolution, and so on - each of these historical transformations had huge and largely unforeseen effects, many of which we regarded as "very undesirable" in hindsight - but the point is, you can't stop these transformations from happening, you can only regulate them, and combat the negative effects.
But, I don't think "the robots (equipped with AI brains) are coming", unless we ourselves allow them to - for the foreseeable future I'm more concerned with what humans with ill intentions, equipped with AI, can do - and my other concern is that the AI scare might distract us from what I think are still bigger threats: the climate catastrophe, and the impending ecological disaster - I see more potential for those wiping us out than for AI to do so.
I totally hear ya! We absolutely need to educate people on the potentials of AI. And I absolutely agree that truth is taking a huge hit by introducing tools like this.
To your points about other potential bigger threats — climate catastrophe and ecological disaster — this is where I get hopeful about AI. I think they mention this somewhere in the video, but I do think AI has the potential to help us combat microplastics and hopefully undo some of the damage that we've done to our ozone layer.
I just hope that (as you mentioned) people with ill intent (or even people trying to do good that cause unintended consequences) don't wipe us all out with AI. I'm not so much thinking about the robots as I am thinking about the unknown... for instance, maybe someone trying to solve the microplastics problem introduces a new compound that is just as harmful, or maybe someone out there with mal-intent creates the next hydrogen bomb level threat.
I see AI as leveling up our abilities on the whole, both to do good and to do harm. Instead of cautiously optimistic, I think I feel hopefully pessimistic.
That's a really good talk!
I genuinely think to much AI will make people more lazy and stupid, that's where the problem really is to me.
Previously I thought AI would wipe humans out by brute force, but never for reasons this subtle.