How Does AI Affect Your Online Shopping?

[Incenjucar]

Yet you're perfectly willing to tell me I haven't seen what I've seen. I was very clear that these were creative works that were then being massaged with AI, then gone over by their producers; they weren't factual works. So the question was whether they changed the meaning and tone, and/or altered the story or scansion involved in a harmful way. They did not.

So I have to conclude you're either telling me I'm lying or incompetent to judge. If there's a third case I'd be happy to hear it.

It will remain in a quantum state. I am not willing to state a firm position on your skill without information, nor am I willing to spend the resources of time, privacy, and pedantic debating of the quality of your work to obtain that information.

The point to be made is that your past employment is not an argument that I find compelling given my current employment. As I said, I am the wrong audience for it.

 

[CleverNickName]

That's quite true. But I'd question how many people are trying for genuinely new creation in the first place. In the vast majority of cases they've been influenced by other images they've seen of things, even if they aren't quite the same thing, and if you're familiar with them, the resemblence will probably be notable. Artists rarely spring forth ex nihilio in their style and topics.

Judging from the number of threads here about an RPG book's artwork, I'd say it's a considerable number. Consumers are fickle, too--an art/music/fashion/film trend that is popular now can be outdated by spring break, but genAI can only push the same variations on the same themes it was trained on. I don't think consumers are going to respond well to entire books of it, over and over again. (But who knows? I just finished saying that consumers are fickle.)

Artists don't spring forth as you say, but they do know how to make something look interesting, creative, and original. GenAI, not so much.

 

[Dannyalcatraz]

The white supremacist tracking example that I linked to, including okstupid.lol and Cybernews, went into more detail on their methods. No facial ID or AI based search results were used.

Ah!

Well, if no AI based search results were used, how would AI facilitate the supremacist search?

The methodology described in the article looks like good investigative journalism connecting the dots of a data breach, consisting partly of self-reported data from people actively concealing their identities to a certain degree.

And yet, as I pointed out, an AI tool misidentified the race of a person who is a low-level celebrity, not concealing anything of the kind. It connected a bunch of public data and still IDed a white person who has been on television as “African-American”.

You do see the problem, right?

 

[Retros_x]

I wonder if they should declare usage of AI if the programmers use it. Because I can guarantee EVERY modern videogame will have source code that was written with help of generative AI.

In general its definitely lessen my want to buy, but its not all gloom and doom. Expedition 33 used AI too and I think no one would describe this game as "AI slop".

I found the question of value change interesting. If I can't perceive the usage of AI the value should not be changed to me.

 

[GrimCo]

Expedition 33 used AI too and I think no one would describe this game as "AI slop".

Had conversation about that topic few days ago with developer friend. In Claire Obscure, devs used AI for placeholder art. Before AI, they would use color gradients, check board textures and basic geometric shapes. It's just one tool to help developers visualize stuff while artist work on real art. AI placeholder assets improve visualization by closely approximating final art fidelity, enabling better judgment of composition, readability, mood, and spatial balance than basic gradients or geometric placeholders. Problem with E33 is that they forgot to replace some placeholders with real assets, but devs forgetting to remove temp assets isn't anything new, that stuff happened before AI also.

In the end of the day, AI is a tool. It's how you use it. Same as hammer. You can build house or you can bash someone's skull.

 

[Morrus]

Give me a break. Lots of us are busy. You made a comment without performing due diligence. It happens to me all the time. Being 'busy' is not an excuse.

Calm down, please.

 

[The Firebird]

And yet, as I pointed out, an AI tool misidentified the race of a person who is a low-level celebrity, not concealing anything of the kind. It connected a bunch of public data and still IDed a white person who has been on television as “African-American”.

I think these are some of the best examples offered. From my perspective, it feels like they don't land because the more pro-AI position is "AI is capable of doing some things well". The counterargument reads to me (have I got it wrong?) as "AI is not capable of doing things well, see these n examples".

And it doesn't convince me because we have all seen many many examples of AI doing things poorly. But those type of examples cannot establish that AI is not capable of doing things well. Just that it often or occasionally does things poorly.

I see the same disconnect in the exchange between @Incenjucar and @Thomas Shey. I'm sure Incenjucar is seeing a lot of bad examples of AI use where it distorts the facts or changes the meaning. But that can't contradict Thomas Shey seeing examples where it doesn't.

If I've mischaracterized the anti-AI argument, then I apologize and would appreciate being corrected.

 

[Belen]

This is an interesting concept. I'm very sympathetic to the goal but at the same time the idea of using an online editor worries me. I'd probably want to write it myself and then upload, which I guess would look suspicious.

The best solution ime is to know the authors and their work and then know if they are trustworthy.

You can upload a word file and have it convert and then edit from there.

Our test authors loved it especially since all the authors can collaboratively edit and there is not need for SharePoint or Google Docs.

It also handles LaTex.

 

[Umbran]

Actually, given the bizarre hallucinations I have seen from LLM, that is both impractical and unethical.

The person who did this had to use a pseudonym and wear a Pink Ranger costume when the hack was done at a conference, because what they were doing was illegal.

So, this is more, "If someone used genAI to break the law and release personal information of private individuals, would you support it?"

 

[Umbran]

I think these are some of the best examples offered. From my perspective, it feels like they don't land because the more pro-AI position is "AI is capable of doing some things well". The counterargument reads to me (have I got it wrong?) as "AI is not capable of doing things well, see these n examples".

And it doesn't convince me because we have all seen many many examples of AI doing things poorly. But those type of examples cannot establish that AI is not capable of doing things well. Just that it often or occasionally does things poorly.

Some quick searching indicates to me that ChatGPT currently hallucinates at a rate of 33% to 79% depending on the type of test.

In cases where you are depending on accurate presentation of information, hallucination is a failure.

Can you claim it does something well, when it fails 33% of the time? I mean, unless you are a baseball player?