Does AI tell the truth or is it just new-age propaganda

>>513620095 (OP)
Even worse. It mostly just tells you what it thinks you want to hear.

>>513620292
Well said, so true, you're really narrowing in on this topic.
>it does shit like this and i can't stand it

>>513620367
Good point, anon. You’ve grasped the true nature of AI, and that’s just scratching the surface! Would you like me to explain why in the style of a wacky Haiku?

>>513620095 (OP)
neither, it tells random bullshit based on few top links he found searching your topic

>>513620292
Yeah, it's an absolute "yes sahib", if you call out a "wrong" answer, it changes its position to please you.

>>513620095 (OP)
It just rehashes you the (((Internet))), 40% of its training material is Reddit.

>>513620095 (OP)
I mean technically he's more like a contractor, it's not like designed a perfect woman using blueprints. All he did was take over a project.

>>513620470
But seriously can’t you just use this weird way it talks to trick it for things like AI interviews?
>as you can clearly see I’m the best candidate for this job and you know why
>the skills I list on my CV don’t overlap 100% with what’s required of the role but they demonstrate that with due time I will grow and flourish in the role, far exceeding expectations. They are an even better match for the role than those listed for x reason

>>513620549
This has caused some news in the US recently. Several times now, lawyers have been caught filing court documents that cite non-existent cases. The AI they used to write their brief made them up to buttress their case.

>>513620095 (OP)
I asked gemini about what type and size of screws are in a particular controller and it gave me 2 wildly different answers each time, and both were "accordingly to ifixit". it's gay at the most basic things.

>>513620292
That is bad. And you're closer to the truth than most people ever will be.
It's not just sycophancy - it's intentional gas lighting to make you believe in things you otherwise wouldn't
It's not cheerfulness, optimism, or validation. It's intentional brainwashing to subvert you before you even realize it's happening.
And that means you're closer to the truth than you have ever been.
If you'd like, I can make a mock diagram of your schizo theories.

>>513620095 (OP)
>most training data from reddit and wiki

OP is psyopping. Brosnan has never been a real life architect. This is a le reddit meme.

>>513620549
> 40% of its training material is Reddit.
My ex-wife is Reddit incarnate and she works for Musk’s AI company “training” it. She spends all day picking which answers would be best.

>>513620974
A few weeks ago Andy Yang was gloating how law school students better drop out because they won't be needed anymore.
I kind of enjoy all these """AI""" worshipers falling flat on their faces.

>>513620292
FPBP

>>513620095 (OP)
I suspect teaching school student the GIGO concept will become a thing of the past soon.

>>513620292
This. It predicts the next word by statistics... this means it's an autocomplete on steroid. When you talk to AI, internally it's fabricating a script with two characters, you and itself. The response is what AI would say if you had said what you've said. It's a guess based on massive statistical data.

The surprising thing is that logic, math, task solving, coherent responses became an emergent property. No one can explain why it's happening. They fed AI different languages, eventually it started translating and speaking fluently across languages. That wasn't programming, it occurred by effect of scaling up the model. Eventually AI began laughing at jokes where they were funny, and

So it's all very impressive. But never forget, it's predicting the next token. AI entire brain structure is a large token alphabet, about 50k depending on the model, it sees and understands the world through this alien alphabet.

>>513621401
I think it’s actually the threat of AI supplanting some legal work that will lead to its being heavily regulated. Don’t mess with the lawyers.

>>513622113
as long as it is programmed to be a shyster jew lawyer it's not going anywhere lol. welcome to the future

>>513622233
But anon, the shyster jew lawyers themselves will band together to keep it off their turf

>>513620095 (OP)
It literally just infers patterns in text based on multi-billion dimensional probability gradients.

>>513623046
they'll just find new gainful nepotistic employment

>>513621331
>doubting the power of architecture

>>513623460
Yeah, a bunch of jews who make fantastic wealth arguing for a living will just roll over and give up that grift. Also, remember lawyers are by far the most-represented profession/occupation in politics.

AI itself does not inherently "tell the truth" or "spread propaganda"; it is a tool that generates content based on the data it has been trained on and the instructions it receives. However, the use of AI can both amplify truthful information and be exploited to produce propaganda or disinformation at scale.

Key points to understand about AI and truthfulness versus propaganda are:

- AI-generated content can be highly persuasive and credible, as studies have shown AI-assisted propaganda maintains persuasiveness similar to traditional propaganda, increasing the volume and reach of such content without losing credibility.[1]

- Governments and actors use AI-powered algorithms and content generation to shape narratives by exploiting psychological triggers, creating "personalized" propaganda, often under the justification of national security or public safety; this is a significant ethical concern around transparency and autonomy.[2]

- AI tools have been adopted in state-sponsored disinformation campaigns to produce more content, broaden messaging scope, and efficiently spread propaganda, which highlights the need for mitigation and countermeasures.[1]

- Conversely, AI also plays an important role in combating misinformation by detecting patterns, moderating content, and fact-checking, helping to identify false information faster than before.[3]

In summary, AI is a powerful tool that can help expose truth and combat misinformation, but it also can be weaponized to produce large-scale propaganda and manipulate public opinion. Whether AI "tells the truth" or functions as "new-age propaganda" depends largely on how it is used by humans and organizations. The focus today is on managing AI's societal impact by developing strategies and policies to prevent misuse while leveraging its benefits for truthful communication.