Gemini illustrates our AI future

These training datasets, where are they coming from?

 

Large public datasets compiled by AI and AI interactions and AI influenced human interactions online.

 

Saw the movie.

Pass.

 

Again, their responses are structured within rulesets, such as linguistic, rhetorical, contextual, positional (refined, aligned or argumentative positions), etc.

Sentience of AI portends the very subversion of humankind. I can envision more than one means of imbuing AI with sentience, one quite simply / affordably, and will not divulge them unless I first see them originate from another source . . . that is the proverbial loaded gun.

 

Me, too. I haven't looked at my vacuum cleaner the same since.

 

I am sending you my dry cleaning bill... I may never get the coffee that just shot out of my nose out of these pants...

 

Yes, but they're not so explicitly defined as to be easily manipulable by programmers. The whole point of them is that by trial and error over billions of iterations they converge on the rulesets of human language on their own.

 

They converge only upon the rulesets of the population with which they interact.

I'll give you an example. I consider myself a pretty good developer of products. A while back LinkedIn began building an AI knowledgebase by soliciting insightful contributions from members in their respective areas of expertise. I first participated, and then realized the end game wasn't public sharing of the information, but aggregating it for viewing only by subscribers of their premium product . . . Full Stop, ToughCOINS!

So, what's that tell you?

First, it tells you that there must be some motivation behind building the AI in the first place, and it's almost certainly not altruistic.

Second, there is motivation behind educating it, not to the benefit of all, but for a targeted audience.

At first, I was willing to share my knowledge. However, once I learned it was only going to serve a limited number of paying members of LinkedIn, and not the entire membership, I ceased contributing. This is my knowledge, not theirs, and I think I'll keep it that way.

Doubtless, the contributing membership does not well-represent all potential contributors, so by limiting its contribution base that AI will converge further from the optimal knowledgebase and content than it could.

 

You're right on both counts, but I'd simply suggest that the goal is quite explicitly to make money, not try to push some ideology.

It reminds me of this take that has slowly and painfully proven accurate with Twitter as old policies slowly work their way back in.



As a recent example, while insisting it's different from old rukes, Twitter just added the policy that purposefully misgendering people is rule-breaking harassment. Unimaginable with the way Musk was talking at first.

Point being, there's not some vast leftist conspiracy behind AI.

Side note, if you sense an ideology being pushed by a company that is out to make money, take that as a sign you're not the target demographic of that company. If it's seemingly every company, take that as a sign you're part of a loud minority rather than the silent majority.

 

While I see your point, the outrage over the Gemini model in particular was that it was not asked to creaet a female Pope or black Viking. I saw a response from Gemini after generating one of these: "Here is a group of Viking warriors with diverse backgrounds" or something of the sort. Now, that seems alright, but the prompt was: "Generate an image of Viking Warriors." The problem with Gemini is that it is inserting this diversity itself, and becoming wildly inaccurate as a result. I unfortunately did not get the chance to try it before they took this feature down, but it did do a good job generating an image of a K-5 Blazer doing a sick burnout.

 

OK... So now I am deviating from the intent of this thread, but I gotta know..... So are you saying I could ask this thing to take my old car and create an image of what I would really like my old car to look like?

 

This Thread has been interesting and fun.

Now, please turn your attention to the UHG debacle, and the incredible impact on our Healthcare System since late Feb., that is about to take down a hell of a lot of Hospitals, Physicians Operations, Pharmacies, etc., due to cashflow payments being disrupted from the majority of health insurance payers....including CMMS Medicare/Medicaid Services.ACA affiliated (or not), holding up BILLIONS of Dollars of daily payments needed to operate. The Breach is across the Board, on the most secure Platform in healthcare, by hackers.
So, AI? Or not?

If you like your internet, you can keep your internet.
No, I am not exaggerating. HHS has now entered the war zone. God help us......

Truth in editing: removed an unnecessary "t", and added a necessary "s". It happens.

 

You bet it can! I can't guarantee the badging will be very accurate, though. Here's a couple I made, you can tell I like old trucks:

 

You would think Justice could have delayed their antitrust probe a couple of weeks to take some pressure off.

 

Definitely, the scary part is that with a bit of reference materiel, anyone can even make convincing photos or even videos of you with your car. You may have an extra finger or two in the photo... but it's getting better rapidly.

It's been interesting to see clearly AI generated photos pop up in cafes here in Korea lately as decorations.

 

Ain't met the vehicle yet that kin put on that kinda smoke show on a dirt road like that'un.

 

Cheby magic

 

. . . an iff'in it could, it'd prob'ly berry itself belly-down fer life.

 

Artificial Intelligence . . . it only helps inasmuch as its programmers are motivated to program the right requirements into it. Failing their imagination of all scenarios, AI is doomed to think more in the box than out of it. That’s true in all disciplines . . . it should come as no surprise that AI is no different.

My opinion . . . The failings of men will only be amplified by artificial intelligence, not eliminated by it.

 

Unless, AI takes out man. Duh.