@manton Maybe enjoy the momentary quiet of the crickets?

Whatever side folks are on hot takes on important questions like AI are messy. Jumping in early with a public take with so many unknowns leaves a lot of room for miscommunication or the sharing of still-in-formation opinions regardless of what "side" anyone is on. It's risky from the start and prone to escalation.

Online conversation in general seems so juiced up in hot takes based on rumors or still-unfolding events.

@dennyhenke That’s a fair point, but maybe the root issue is the “sides” to begin with. I took a look at the facts of what we knew and made a call. But that’s good advice to enjoy the quiet! 🙂

@manton Absolutely agreed on the root issue of sides. Imagine a world where we humans were better at working together. Where the goal was always improvement of the collective good enabled by always seeking truth rather than being right.

On every level, from the personal to the international, we waste so much time, energy and resources being stuck in relations based on seeking dominance and power-over.

For all our "intelligence" we've still not figured out how to work together.

@dennyhenke Well said.

I am still with you, for the record. We don’t agree on everything, but that’s exactly why I continue to read your writing. Thank you for sharing your thoughts, and thank you for creating Micro.blog, which is still my preferred way to follow these kinds of conversations.

@fgtech Thank you. Totally fine of course not to agree with everything!

This new twist only shows that OpenAI was hiring voice actors back in May 2023. While the company may not have explicitly requested a voice that sounded like ScarJo, it does not say anything about whether Altman weighed in on the selection to ensure that the outcome was Her-like.

@manton Maybe I was misreading other takes, but I was never under the impression they used a synthetic *copy of her voice*; I still thought it was yet another tasteless, legally questionable (“likeness” rights, etc.) blunder clearly posturing to be as “Her”-like as they could get away with. In the same vein as so much else they’ve built with a “we’ll do/take whatever we want” attitude, unfortunately.

Always enjoy reading your thoughts, hope you’re able to quickly brush off this week’s noise!

@cristian You’re moving the goalposts! 🙂 I will concede that Sam is fascinated with Her and probably hoped to get a voice that felt familiar, but that is a very different thing than “she declined so they copied her voice”.

@matt_garber Thanks, trying to brush it off! I think there were a range of opinions, from an actual “clone” of the voice, to a sort of “inspired by”. The phrase I used was whether they were “ripping off” her voice, which is up for interpretation too, I guess.

@dennyhenke if there was one thing I wish people could internalize, it would be this. I was trying to get at something similar in my post on Trump this morning.

I’m moving the goal posts? LOL. Point taken, though. But it does seem odd that he would go asking her to use her voice when they already had a voice that sounded like her. Why?

@manton It’s not like Apple & Google & Microsoft included it in their job listings back when they were colluding to depress salaries. The fact that they didn’t explicitly call for a ScarJo voice doesn’t mean it didn’t happen. “Altman was traveling a lot then” is pretty weak proof.

Also, what they set out to do is probably less relevant than what they delivered. Even if just Altman thought “oh hey this coincidentally sounds like her let’s leverage that” she’s got publicity rights protection.

@cristian My guess is Sam thought it would be cool to have an additional voice that was identical, maybe even called Samantha. It would become the default and the 5 other voices would be options. Meanwhile the team pushed ahead with the voices in progress.

@donw It does seem like there might be a legal case, even if it’s not as bad as people were assuming.

I am not convinced this was coincidental. And I think that OpenAI planted the story in WaPo (as someone who works in PR and was a former journalist) as a first step to change the narrative because of the high legal risk. “Documents shared by OpenAI”? Not a coincidence. This is crisis communications at work.

I feel bad that you felt bad, but that “Crickets” comment really does not help. People may not have engaged directly with you, but I don’t see this as a resolved case now that the truth™ is out as you seem to.

Look for example here: news.ycombinator.com/item That discussion doesn’t sound like crickets to me.

@matti I think it’s reasonable for Manton to feel like it’s “crickets” in the sense that he was directly criticized for suggesting the events might look at lot like they did. Those people didn’t re-engage or reconsider at all. That’s not to say that what OpenAI did was right or moral (I personally don’t think it was). It is to say the possibilities folks thought Manton was immoral/unethical for even entertaining turned out to be quite possible.

@jsonbecker I guess you have a point there. As somebody who also was part of this (but more from the perspective of being critical of not including the aspect of power relations and track records into these reflections - or at least that was what I was thinking when commenting originally) it seems hard not to read this at least partly as vindictive, though. And I think it’s a little early for that. I also suspect it’s the “crickets” that spurred others to comment again critically (just like me, I admit).

Assuming the story isn’t just PR from OpenAI, which is what it appears to be, they are claiming it was done out of incompetence, not malice. This does not help their case.

@matti As one who did originally engage and didn’t, until now, with that “news”, I didn’t mostly ouf of exasperation. I didn’t see a point in continuing, Sides have been chosen.

@matti I agree this story could flip again if more facts are revealed. I don’t think I’m going to say more other than this whole thing is disappointing to me, and I’m looking forward to moving on and writing some code.

@mcg You are putting “news” in quotes from the Washington Post? OpenAI obviously fed them the sources for that story, but a journalist did work on it and interview people. Thinking about sides is the problem. I think I’ve been pretty fair to all aspects of this story.