TVTropes Now available in the app store!
Open

Follow TV Tropes

Following

Artificial Intelligence discussion

Go To

With how much artificial intelligence has been improving, in many areas such as text reading/generation, picture reading, picture generation, convincing voice synthesis and more, I think there's a lot that can be discussed, about the effects that this technology will have on society.

I'll start off with one example.

I'd been thinking about the enshittification cycle of tech, and I think it's coming for Google hard. The search engine just isn't so great at finding what you actually want, and I think that's gonna leave a big opening for Bing with their use of AI. If the AI can sift through the crap and actually find what you want for real, due to its understanding of language, it'll actually make searching super useful again.

In the pre-Google internet, search engines used to search only for exact words and phrases, which had its uses, but also meant finding a lot of sites that simply crammed in a lot of popular words and phrases to get visitors. Google cut through the crap with a better understanding of how to "rank" sites relative to how relevant they are, and even find sites that are on the topic you were looking for without using the same exact words.

But Google started to become more advertiser-friendly, then later, more shareholder-friendly. There's a limit to how much one can make their product built entirely around shareholder growth, so as it turns to crap, it leaves an opening for a competitor to show up.

Since Bing/ChatGPT (which Bing is plugged into now) understands the use of language, it can actually understand context and determine relevance based on that. And that'll make it huge, I think. Context-based understanding of web pages can potentially do an excellent job of finding what people actually want, in a way that goes way beyond Google's page ranking systems, or the examination of exact words.


There is also a more specific thread for the legality, ethics and nature of AI art. Posts on this aren't off-topic here, but may sometimes be more appropriate to the other thread.

Edited by Mrph1 on Jun 22nd 2024 at 11:53:33 AM

HavocCrow Havel's apostrophe from tower in the fountain of sparks (Plucky Ensign)
Havel's apostrophe
#1376: Jan 31st 2025 at 4:26:51 AM

@Big Bad Shadow 25

I do remember hearing about a chatbot that plugged into social media and developed into a Nazi AI from all the racism.

You may be talking about Tay, a Microsoft-developed bot which posted on Twitter and supposedly started to post racist Neo-Nazi tweets.

(...except that the story is dubious at best, and might originate from a few trolls abusing the bot's 'repeat after me' function to make it say skeevy stuff.)

BigBadShadow25 With my last breath, I curse Zoidberg. from Basement at the Alamo (Experienced, Not Yet Jaded) Relationship Status: Desperate
With my last breath, I curse Zoidberg.
#1377: Jan 31st 2025 at 5:38:42 AM

[up]Yeah, that was it.


Basically I think you are getting anxiety over scifi concepts that require multiple steps to jump over before they'd be anywhere realistic to happen. Lot of singularity super ai hype works on same basis as religion. Its belief in supreme being and faith that it can exist.

Yeah, the way everyone talks here makes me wonder how quickly those steps will be gone over.

You’re Gonna Carry That Weight.
Cid El Cid Since: Jul, 2015 Relationship Status: Hiding
El Cid
#1378: Jan 31st 2025 at 6:00:23 AM

Some news.

Goggle's Gemini 2.0 Flash has been 'officially' released after the experimental version was made available on December 11th. The old versions, 1.5 Flash and 1.5 Pro will still be available for a few weeks.

There's still no news on when the 2.0 Pro version will officially launch.

I did some brief testing with all three versions, 1.5 Pro, 2.0 Flash, and 2.0 Advanced, and Flash is still behind the other two for more complex tasks. I can't wait to have the official release of 2.0 Pro so I can use its real-time features.

Edited by Cid on Jan 31st 2025 at 9:05:54 AM

Fighteer Lost in Space from The Time Vortex (Time Abyss) Relationship Status: TV Tropes ruined my love life
Lost in Space
#1379: Jan 31st 2025 at 6:01:17 AM

Nobody here, as far as I know, is an expert in AI development, so we're all just spitballing based on public information. Don't take any of this too seriously, or at least make sure to check the conversation against actual reality every now and then.

Personally, I don't think we're about to be taken over by our robot masters any time soon, but it is interesting to speculate about what such a world might look like. My motivation for participating is that I am extremely skeptical of any claim that it would be a Skynet situation.

"It's Occam's Shuriken! If the answer is elusive, never rule out ninjas!"
BigBadShadow25 With my last breath, I curse Zoidberg. from Basement at the Alamo (Experienced, Not Yet Jaded) Relationship Status: Desperate
With my last breath, I curse Zoidberg.
#1380: Jan 31st 2025 at 6:01:49 AM

Alright. Let’s be sure we’re clear about that in the future.

You’re Gonna Carry That Weight.
SpookyMask Since: Jan, 2011
#1381: Jan 31st 2025 at 7:41:09 AM

Its basically same as back when people were optimistic about flying cars in future. Sure tech development speed is fast, but most future visions aren't that accurate especially when it comes to people making bold proclamations on exact time frame(like, we know that eventually we will run out of fossil fuels at some point, but by some old predictions we already would have run out of them). So it comes down to speculation being fun and people like dreaming about what future could be like.

Marchen Too hot (Hot damn) from Somewhere Out in Space (Experienced, Not Yet Jaded) Relationship Status: It's not my fault I'm not popular!
Too hot (Hot damn)
#1382: Jan 31st 2025 at 7:56:49 AM

[up] I assume that's not intentional, but I'm suddenly reminded of the fact that flying cars turned out to be really impractical for any real-life purpose. Like, not a ton of reasons to have a car that's also a plane, when you could just use a plane instead. I assume AI would have more practical applications.

Read Otr of the Flame (She/Her)
megarockman from The Sixth Borough (Experienced Trainee)
#1383: Jan 31st 2025 at 8:02:23 AM

I would think AI would also be easier to specialize as necessary for given tasks. If nothing else, forking an offspring if it turns out some area is better to be concentrated on, while perhaps structuring in at least an ability to have "siblings" talk with each other to share data if going cross-discipline might yield new suggestions.

The damned queen and the relentless knight.
Silasw A procrastination in of itself from A handcart to hell (4 Score & 7 Years Ago) Relationship Status: And they all lived happily ever after <3
A procrastination in of itself
#1384: Jan 31st 2025 at 12:15:04 PM

What if the profit-maximizing AI tells its corporate masters to be ethical citizens and stop being so greedy because it understands the tragedy of the commons?

So I had a discussion with a colleague about something similar. What happens if we train our AI to follow our stated ethics and laws? Most large companies have statments about their values and will insist in all formal documentation that they follow the law.

So if you’re programming a cooperation managing AI my theory is that you’re liable to program it based on such statments. Maybe because you legally have to (governments requiring AI making hiring decisions to be program with knowledge of labour laws is easy to imagine), maybe because your top leadership are self-deluded enough to think they are following said values even when they’re not (I suspect most of us have had a boss like that), maybe because the programmer doesn’t want their name coming out in a future court case as the person who told the AI to break the law for max profit.

That means the AI could end up following both the law and cooperate ethnically principles better than humans do, which I think will see a lot of said AI’s being unplugged much the way Air Canda went back to humans after the chatbot started issuing refunds they didn’t want to pay out.

In the end you can make an unethical AI, but you have to either be honest with it that that’s what you want it to do (hard with the human tenancy for self-delusion), avoid any training on ethics at all (viable with small A.I.s but something that will probably be mandated for big ones) or train it not via programming but by having it shadow and mirror an unethical human (something that would take time and may not work if the human knows the AI is monitoring them).

“And the Bunny nails it!” ~ Gabrael “If the UN can get through a day without everyone strangling everyone else so can we.” ~ Cyran
Fighteer Lost in Space from The Time Vortex (Time Abyss) Relationship Status: TV Tropes ruined my love life
Lost in Space
#1385: Jan 31st 2025 at 12:39:09 PM

[up] That reminds me of a challenge that Tesla faced when training its Full Self-Driving software. It used real driving behavior, but real drivers break laws all the time, such as the requirement to stop at stop signs. To be compliant with the laws (as demanded by regulators), they had to put in some hard tweaks to the training rules so that the software would go along with the five percent or so of data showing full stops.

Then customers got angry because they hated it when the cars made full stops at stop signs.

Worth noting is that this only occurred when there was no contesting traffic. Obviously, the car always stops if there are other vehicles at the intersection. But most US drivers approaching an empty intersection will roll the stop sign.

AI learning models need to comply with ethics and laws, for obvious reasons, but they also need to be usable in real life, and if customers won't use the software unless it behaves in ways that are illegal and/or unethical, a dilemma arises.

One might observe that laws requiring full stops at empty intersections are silly and/or impractical, but that doesn't solve the problem. The laws exist and are enforced, and no regulator will approve software that deliberately breaks them.

ETA: I assume the conflict has been resolved, since I haven't heard about it in a long time. ETA 2: A quick search reminds me that FSD was updated to always stops at stop signs regardless of user preferences, but nothing since then.

Edited by Fighteer on Jan 31st 2025 at 4:06:45 AM

"It's Occam's Shuriken! If the answer is elusive, never rule out ninjas!"
Silasw A procrastination in of itself from A handcart to hell (4 Score & 7 Years Ago) Relationship Status: And they all lived happily ever after <3
A procrastination in of itself
#1386: Jan 31st 2025 at 2:13:08 PM

Another one I know I’ve seen is indicating when changing lanes, my mother briefly had a courtesy car with lane assist and she hated that it kept pulling against her when she was changing lanes, until I pointed out that she was failing to indicate and that’s why the car thought she was drifting instead of making a deliberate manoeuvre.

In general I suspect that people will care less about such things once self-driving cars are actually self-driving the couple seconds lost by not rolling a stop sign are annoying when you’re focused on the road, but they’ll be unnoticed when you’re reading a book.

Edited by Silasw on Jan 31st 2025 at 10:15:06 AM

“And the Bunny nails it!” ~ Gabrael “If the UN can get through a day without everyone strangling everyone else so can we.” ~ Cyran
Kayeka (4 Score & 7 Years Ago)
#1387: Jan 31st 2025 at 2:29:33 PM

[up]Are we ever going to get that, though? Because if we get to the point where producers can sell cars with Self Driving For Realsies This Time, we've also reached the point where said producers are going to have to assume a certain degree of responsibility over any accidents said cars might cause.

That's going to be a lot of legal fees.

Silasw A procrastination in of itself from A handcart to hell (4 Score & 7 Years Ago) Relationship Status: And they all lived happily ever after <3
A procrastination in of itself
#1388: Jan 31st 2025 at 2:46:53 PM

Mercedes have publicly said they will take legal liability for accidents involving their lv3 self-driving system if it’s used in the situations it’s authorised for (highway/motorway driving).

“And the Bunny nails it!” ~ Gabrael “If the UN can get through a day without everyone strangling everyone else so can we.” ~ Cyran
Kayeka (4 Score & 7 Years Ago)
#1389: Jan 31st 2025 at 3:08:08 PM

[up]So quite pointedly, not any situations involving stop signs, I believe.

Fighteer Lost in Space from The Time Vortex (Time Abyss) Relationship Status: TV Tropes ruined my love life
Lost in Space
#1390: Jan 31st 2025 at 3:13:47 PM

Waymo and Cruise already operate "fully autonomous" vehicles and accept liability for them. This isn't a weird or unforeseen situation.

"It's Occam's Shuriken! If the answer is elusive, never rule out ninjas!"
Kayeka (4 Score & 7 Years Ago)
#1391: Jan 31st 2025 at 3:37:55 PM

And no "the system was disengaged .5 seconds before impact" loopholes? Well, I suppose that's one concern less.

Fighteer Lost in Space from The Time Vortex (Time Abyss) Relationship Status: TV Tropes ruined my love life
Lost in Space
#1392: Jan 31st 2025 at 4:22:19 PM

That "loophole", as you call it, is irrelevant for fully autonomous vehicles as they will be in control 100% of the time. The Tesla Cybercab, for example, won't even have driver controls. Unless the passengers actively vandalize it, there would be no way for them to assume any liability for crashes.

Edited by Fighteer on Jan 31st 2025 at 9:10:37 AM

"It's Occam's Shuriken! If the answer is elusive, never rule out ninjas!"
BigBadShadow25 With my last breath, I curse Zoidberg. from Basement at the Alamo (Experienced, Not Yet Jaded) Relationship Status: Desperate
With my last breath, I curse Zoidberg.
#1393: Jan 31st 2025 at 5:39:02 PM

Honestly, given how I’ve come to dislike driving in recent years, I’d… probably be okay with automated cars.

Honestly, it’s the traffic I can’t stand. Most of the time, not even caused by an accident.

You’re Gonna Carry That Weight.
Protagonist506 from Oregon Since: Dec, 2013 Relationship Status: Chocolate!
#1394: Jan 31st 2025 at 5:45:30 PM

I strongly suspect that driverless vehicles without even the option to drive yourself feels like a way off.

Leviticus 19:34
Kaiseror Since: Jul, 2016
#1395: Jan 31st 2025 at 6:02:51 PM

I'm a bit wary about self-driving cars. Mainly because, as mentioned above, human drivers tend to break the rules alot, whereas a self-driving car would follow them. I'd imagine even as self-driving cars become more accessible, there would be many people not using them who'd be driving like assholes.

Fighteer Lost in Space from The Time Vortex (Time Abyss) Relationship Status: TV Tropes ruined my love life
Lost in Space
#1396: Jan 31st 2025 at 6:12:07 PM

We have a self-driving vehicles thread. I suggest we take the detailed discussion there. I brought it up as an example of a situation in which customary behavior is not aligned with "lawful" behavior, and thus an AI trained to be lawful will run afoul of custom.

FYI, Tesla is already developing the Cybercab. The day is closer than you think. While people are busily debating its merits, the software is getting really good.

Edited by Fighteer on Jan 31st 2025 at 9:13:23 AM

"It's Occam's Shuriken! If the answer is elusive, never rule out ninjas!"
Cid El Cid Since: Jul, 2015 Relationship Status: Hiding
El Cid
#1397: Feb 5th 2025 at 7:07:19 AM

Google's Gemini has released versions 2.0 Flash Thinking Experimental and 2.0 Pro Experimental. So far it seems pretty cool and its answers are better than the old experimental model. Or maybe I am hallucinating, hehe.

Also, a couple of days ago they announced an update for Imagen 3, Google's image generation AI, and they claim it's ranked #1. I don't use it for work so don't know if their claim is true or not, but from what I've read, Imagen is really good.

BigBadShadow25 With my last breath, I curse Zoidberg. from Basement at the Alamo (Experienced, Not Yet Jaded) Relationship Status: Desperate
With my last breath, I curse Zoidberg.
#1399: Feb 7th 2025 at 12:44:22 PM

I thought this article from Futurism was worth sharing. It’s about how men are creating AI girlfriends so they can verbally and emotionally abuse them. https://futurism.com/chatbot-abuse

https://static.tvtropes.org/pmwiki/pub/images/image_1_1.jpeg

Edited by BigBadShadow25 on Feb 7th 2025 at 3:44:45 PM

You’re Gonna Carry That Weight.
Kaiseror Since: Jul, 2016
#1400: Feb 7th 2025 at 2:38:57 PM

[up] Wasn't expecting the robot apocalypse to start like that...


Total posts: 1,707
Top