Hacker News Reader

Comment by Entinel | Hacker News Reader

This is pretty naive. My fear isn't that AI is going to suddenly turn into Skynet(although I think it is possible) but that AI is going to:

1. Enable more sophisticated scams and exploitation.

2. Create even more distrust in people through the use of Deepfakes and AI generated voice.

3. Put a lot of people out of work without the required safety nets to give them time to find their way through the turmoil.

4. Lead to a giant deluge of content spam that is just regurgitated AI text and art that makes the discoverability problem even harder to solve than it is now.

And from what I can tell, all 4 of those are currently happening.

View on HN

Skynet aside, aren’t all these other fears also just fear of the future too?

Cool, so any consideration of the downsides of technology is just "fear of the future"? We're just suppose to terminate all thinking and mindlessly charge ahead?

Why would we ignore it? The purpose of fear is so we do give it more consideration and don’t charge ahead.

I don't see how that's a useful discussion though. It's not really lofty futurist fearmongering, it's uncertainty about a very real, currently happening event in history. It would be like dismissing someone in the US for worrying about WW2 in 1939. Yeah, it might not impact the US, but also it's not immaterial daydreaming either. It is something you can reason about and prepare some plans for.

Sure, if speaking generally. I mainly take issue with the author passing it off as fear of change or fear of the worst case scenario as if it were just a phobia. The things I listed I feel are very real and actively happening.

It's already happening. Numerous deepfakes have spread around online, and possibly many more that nobody even realizes are deepfakes. There are numerous stories of people losing their jobs because ChatGPT. One story comes to mind, a freelance writer who lost their biggest client because they started using ChatGPT to generate everything.

It's already happening now. It's only going to get worse as these AIs improve.

a spear flying at your face is fearing what will happen when it hits you in the face.

all fears are fear of the future when you apply enough abstraction.

These sound no different to me than any other transformative tech during other time frames in history. More scams [some people eventually learned, those who didn't lost out], less of the same jobs we have now [different jobs were created, those who didn't adapt or refused lost out], more noise [people learned to filter differently, those who didnt or refused lost out].

Is talking about the potential impact of climate change on people around the world fear of the future?

I think the bigger idea here that hanson maybe doesn't do a good job of highlighting, using your example, is that if you look at the historical examples (ice cores, old rocks, etc) of big shifts in the world climate, it's going to happen regardless of humans.

This sounds awfully handwavy. The fact that the climate changes over millennea doesn't do anything to justify climate destruction. For a small scale experiment, try cutting down all of the trees in your neighbor's yard and justifying it with "Well all trees fall down eventually".

In your example, the likely outcome depending on the jurisdiction would be some kind of fine (neglecting some kind of physical confrontation). Just because something negative would happen to me (maybe more parties as well if one includes the trees, the neighbor, anyone/thing else that tries to be involved in such), doesn't really change the longer scale. We kill plenty of things in our daily lives without a second thought with every breath we take, to favor some killing over others is arbitrary. Some will manage to survive in some form (maybe even unrecognizable to humans today) is such an end state where all trees are gone, most who are dogmatic about others trying to do so, will not.

Many eukaryote's from billions of years ago died out, and some went on to become humans. If one could imagine that those eukaryote's that died out then could ponder the future, no doubt plenty would have sought to be comforted by some kind of pause in their development for some kind of eukaryote safety net to avoid such far future couplings in the form of humans.

> If one could imagine that those eukaryote's that died out then could ponder the future, no doubt plenty would have sought to be comforted by some kind of pause in their development for some kind of eukaryote safety net to avoid such far future couplings in the form of humans.

That's the kind of short term thinking that gets eukaryotes in trouble. The heat death of the universe in 1.7e106 years means none of that ultimately matters, right?

> The heat death of the universe in 1.7e106 years means none of that ultimately matters, right?

Makes no sense for an actor that can perceive such a theory among many others that they cannot currently test, to weight that higher for any other (esp ones that they can test today despite others telling them not to, who are without any feasible means at scale to stop such).

Yes. We’re afraid of what it may do to those people in the future.

You’re generalizing the particular. This is a fear of a very specific variety of future. Nobody is afraid of the a future that has a vaccine for HIV or the elimination of extreme poverty. The one where we are inundated with AI is special.

About the AI art aspect, having just experienced this week what's probably the greatest hour of TV ever filmed until now, certainly on a personal level, in Succession's episode 3, season 4, I am fairly certain AI tools without agency will never be able to achieve such levels because art is fundamentally about humans, the manner of depiction is secondary: art has been impressive from Lascaux to Giotto to Stable Diffusion, the point however is the transference of feelings, from the artist to the viewer, and for transference to be achieved the artist must have feelings to begin with.

This is not to say that great artists of the future won't use AI tools for shaping their vision. And if AI goes beyond being a mere transformer of input and starts having agency, well, we're not there yet at least.

> artist must have feelings to begin with

This is an interesting concept since researchers really aren’t sure how consciousness in humans and animals even happens. We just don’t know.

Feelings could be theoretically be possible in artificial intelligence. We really don’t know enough about the subject to say either way.

Sure, "feeling" is rather ineffable and plenty of ink and bits have flown trying to fixate it. Taking it down a notch, you can probably put it as a very strong dichotomy like-dislike. I know my dog has some proto-feelings because he really likes belly rubs and he really dislikes pulling his tail. Nevertheless, there is probably something akin to the Weber–Fechner law [1] through which "feelings" will be disambiguated. And when that happens, we might discover it was us who have been artificial from the start: biochemistry being merely ridiculously advanced technology, indistinguishable from magic as the saying goes, by comparison with our rusty transistors.

[1] https://en.wikipedia.org/wiki/Weber%E2%80%93Fechner_law

To add to this, the fact that AI tools are black boxes that not even the creators can fully understand. This leads to a "trust us" mentality where actual answers about why something is can never actually be answered.

The TV show Humans is what, 2 years old now, and one of the main characters was worried they were not needed by the future because robots would take all the jobs.

Now IRL, there already is a steady stream of anxiety & angst, regular questions of "why would I bother becoming a coder anymore, since all the jobs will be wiped out by AI?"

To me it's an indicator we overly mythologize these things we cannot know. We not just believe the hype, but we amplify it enormously.

After Zuboff's Age of Survelliance Capitalism, Cory Doctorow's put out a response position that while the book is great, it 100% accepts every capability these folks say they have to completely reprogram human behavior, and it takes it further.

This tendency to assume technologies have high oracular powers, to- even when we are critiquing (Zuboff)- assume the system is at least if not vastly impactful & powerful than the claims... it's a savage problem for society, a nature of people & how we view the unobservable unexplained world as mystical & overaweing. How we dampen the hysteria is beyond my guessing, atm.

>3. Put a lot of people out of work without the required safety nets to give them time to find their way through the turmoil.

I agree with this, but I also think the risk is less that lots of people will be "successfully" replaced by AI systems, and more that management will force the adoption of models as cost cutting measures. I wrote at some point that people shouldn't be afraid that AI will do their job, but that managers will think that it can because they don't understand the exigencies of the job or the limitations of the model.

Yeah I think we'll see a lot of Good Enough (to cut salary expense) replacements rolled out, which is even more disturbing. The middle class will shrink and in exchange we'll get crappier products.

To your point #1. The first conviction for printing counterfeit money happened only 3 months after the introduction of the US dollar. Scams always follow new technology. I see no reason why now should be different.

It will likely balance out as ML/AI will also provide more sophisticated tools to deal with scams, exploitation, spam and fakes.

It will be a constant arms race though and I don't see how the detection side could possibly keep up. The generative models will continue evolving and propogating. Any detectable pattern in one model won't be reliable for other cases.

What jobs are currently being displaced?

Not about to be but currently are…

You can find stories online about content writers and what not being replaced but ones that I personally know of are a lot of minor art roles in game development being replaced. Things like icon art for example that would usually be outsourced to a contracting studio. I will give the disclaimer that I don't work in game development but I'm in the PNW so I know a lot of people who work at studios.

What sort of numbers are we talking about?

> a giant deluge of content spam

This is a small tweak in a search engine's algo that /detects/ AI and downranks it. Only organic human-crafted content should be allowed to surface in a search engine.

You’re eliding how hard that detection bit is. As of this moment, it’s effectively not possible to do accurately, and there’s nothing visible on the horizon.

You really believe we can create AI-detection algorithms that will stay ahead of the content generators?

Reply on news.ycombinator.com