The Scarlett Johansson Incident Makes OpenAI Look Desperate | Hacker News Reader

The Scarlett Johansson Incident Makes OpenAI Look Desperate

nymag.com

62 pointsjbegley17 comments

I’m sad this is the outcome. “Her” was a spectacular and novel exploration of how intelligence fundamentally drives us towards curiosity & connection. It also has a stunningly unusual ending given the topic, where the human protagonist grows in their humanity instead of getting lost in a digital simulation of it.

Johansson is a tremendous artist & entrepreneur, her decision is probably the most reasonable one at this point in AI’s development. Given the apocalypse AI will likely cause to the business model of the creative industry, any AAA celebrity probably needs to avoid being the “identity of AI.”

OpenAI’s and Sam’s enthusiasm for the idea makes a ton of sense. If you’re a tech optimist and haven’t seen “Her,” it’s worth the watch.

>Given the apocalypse AI will likely cause to the business model of the creative industry

We're three years into AI being "capable" of creating more or less realistic looking images, pieces of art, text content and now also video, and I still see no apocalypse. Creatives haven't lost their jobs, many writers I know haven't lost theirs either and both photographers and other visual artists continue to work, sell and hold expositions. At least in the otherwise largely unrelated worlds of fine art appreciation and commercial content creation, it seems that few people take AI sludge seriously so far or trust it much do do anything important, especially now that the initial hype bled down and revealed a lot of tepid flaws.

On the other hand, for industries where visual designers were used to create quick commercial graphics, possibly yes. Many business owners and marketers do use AI there as far as I can see, because they only need a quick picture and don't care much about other details.

Also, don't know the deeper details of her conflict with OpenAI, but good for Johansson for publicly slapping this generally shitty, grossly hypocritical company down a notch. OpenAI as a business needs plenty of that.

It seems you think three years should be long enough for you see the apocalypse. Why is that enough time, do you think? How long was the web around before it caught fire and changed everything?

> AI companies have a funny relationship with sci-fi.

I feel this ties into another recent submission "For tech CEOs, the dystopia is the point" [0], which refers to some of the same tweets/similarities.

[0] https://news.ycombinator.com/item?id=40371835

Who could be surprised? All of OpenAI's success turn on them taking without permission and/or legal right.

Isn't it just a PR stunt?

They get to use her voice for some time, generating hype. Then get even more coverage over the controversy, for free, while talking about their new product.

Not desperate at all.

This is modern marketing.

It’s certainly a strange incident but I’m not sure about desperate.

If calling to ask again 2 days prior to release doesn't reek of desperate, then what does?

"We're going to release not-your-voice in 2 days anyway, maybe you want think about the deal again?"

Do you think that's what happened given that they then removed it?

They released this voice like 6 months ago and no one noticed.

One billionaire and one millionaire arguing over whether or not creating a news cycle with both of their names in it somehow constitutes theft or damages.

Literally couldn't care less.

TL; DR “You’re either the company harnessing the barely controlled phenomenon of imminent self-replicating machine intelligence, leading humanity into its next technological epoch, or you’re a mid-stage start-up that for some reason really needs to copy that voice from that movie to market an incremental product upgrade.”

It also shifted my perception from “omniscient super app close to AGI” to “horny waifu assistant”.

With all of that Microsoft funding, they couldn't just pay Jen Taylor?

GPT-4o is a phenomenal leap forward for true multimodal models, inference efficiency, and speech generation; it's also replacing the free public model. That's massive news. In contrast, the Sky voice drama is vastly overblown and a blatantly obvious/predictable/avoidable mistake by OpenAI. But it'll be memory-holed in a week. Two, at most.

ClopenAI is a terrible company, but calling them desperate over a PR faux pas is pathetic.

They had to have known that this would be brought to light when releasing something that they didn't get permission for. Asking twice means that they knew it sounded very like SJ. My theory is that this will be used as an example for why legislation needs to be pushed harder with tighter controls and official hardware/software locks. "We can't allow the masses access to generate whatever they want."

Reply on news.ycombinator.com