Regardless of the exactly voice spectrum, the plot would apply with any flirty female voice. It was not a movie about Scarlett Johansson. It was a movie about AI eliciting a relationship.
For the “her” reference(s?), was there anything beyond the single tweet?
100%. This whole thing is more stupidity than anything else. There is nothing wrong with using a voice that sounds like her. There is everything wrong with referencing the movie and sort of implying it is the voice from the movie. They could have easily let others make the connection. So dumb.
Not an IP lawyer, but I think the company that produced the movie owns the relevant IP, and Johansson might also own IP around it.
You can have an opinion on it, but they are going to get sued. Just like I can't take Moana and throw her in an ad where it says "I like [insert cereal here]", they can't take a character and use it without expecting Disney/whoever to come sue them.
Hmm. Being able to say "thou shalt not make a character similar to Her" is a lot like saying "thou shalt not make a video game character similar to any other." It’s not an explicit copy, and their name for Sky was different. That’s the bar for the videogame industry; why should it be different for actors? Especially one that didn’t show her face.
This whole thing is reminiscent of Valve threatening to sue S2 for allegedly making a similar character. Unsurprisingly, the threats went nowhere.
It’s the other way around. The contortionists are on the other side of the issue. We’re talking about OpenAI hiring someone to use their natural speaking voice. As movies say, any similarity to existing people is completely coincidental from a legal perspective.
From a moral perspective, I can’t believe that people are trying to argue that someone’s voice should be protected under law. But that’s a personal opinion.
They said so, and it’s what I would have done. I have no reason not to believe them.
Unfortunately a commenter pointed out that there’s legal precedent for protecting people’s voices from commercial usage specifically (thanks to a court case from four decades ago), so I probably wouldn’t have tried this. The cost of battling it out in the legal system is outweighed by the coolness factor of replicating Her. I personally feel it’s a battle worth winning, since it’s bogus that they have to worry about some annoyed celebrity, and your personal freedoms aren’t being trodden on in this case. But I can see why OpenAI would back down.
Now, if some company was e.g. trying to commercialize everybody’s voices at scale, this would be a different conversation. That should obviously not be allowed. But replicating a culturally significant voice is one of the coolest aspects of AI (have you seen those recreations of historical voices from other languages translated into English? If not, you’re missing out) but that’s not what OpenAI did here.
No. But in this particular case, there are two factors that make that irrelevant for me. One, I would have made their same mistake. (If I was Sam, I too would have found it a really cool idea to make GPT have the voice of Her, and I too would not have realized there was one dumb court case from the 80s standing in the way of that.)
Two, it’s bogus that conceptually this isn’t allowed. I’m already anti-IP — I think that IP is a tool that corporations wield to prevent us from using "their" ideas, not to protect us from being exploited as workers. And now this is yet another thing we’re Not Allowed To Do. Great, that sounds like a wonderful world, just peachy. Next time maybe we’ll stop people from monetizing the act of having fun at all, and then the circle of restrictions will be complete.
Or, another way of putting it: poor Scarlett, whatever will she do? Her voice is being actively exploited by a corporation. Oh no.
In reality, she’s rich, powerful, and will be absolutely fine. She’d get over it. The sole reason that she’s being allowed to act like a bully is because the law allows her to (just barely, in this case, but there is one legal precedent) and everyone happens to hate or fear OpenAI, so people love rooting for their downfall and calling Sam an evil sociopath.
Someone, please, make me a moral, ethical argument why what they did here was wrong. I’m happy to change my mind on this. Name one good reason that they shouldn’t be allowed to replicate Her. It would’ve been cool as fuck, and sometimes it feels like I’m the only one who thinks so, other than OpenAI.
Actually, there's a similar court case from 1988 that creates legal precedent for her to sue.
"That's just one case! And it's from 1988! That's 36 years ago: rounded up, that's 4 decades!"
Actually, there's a court case from 1992 that built on that judgement and expanded it to define a specific kind of tort.
"That's bad law! Forget the law! I demand a moral justification."
Anyway, asking a person if you can make money off their identity, them saying no, and you going ahead and doing that anyway seems challenging to justify on moral grounds. I don't think you're willing to change your mind, your claim notwithstanding.
If you approach a debate from a bad faith standpoint, don’t be surprised when the other person doesn’t change their mind. "I think you’re a liar" is a great way to make them nope out.
Which is a shame, since you had a decent argument.
Except it isn’t. Again, you’re acting like OpenAI tried to profit off of Scarlett. They tried to profit off of the portrayal she did in the movie Her. These are not the same thing, and treating them as interchangeable is some next level moral rationalization. One is taking advantage of someone. The other is what the movie industry is for.
Now, where’s this case from 1992 that expended and defined the scope of this?
> Except it isn’t. Again, you’re acting like OpenAI tried to profit off of Scarlett. They tried to profit off of the portrayal she did in the movie Her.
Ahhh... so you admit OpenAI has been shady, but you argue they're actually ripping of Spike Jones not Scarlett Johansson?
HEH. The people who say Sam is shady aren't really interested in this distinction.
(And you're wrong, both ScarJo and the film own aspects of the character they created together.)
> Again, you’re acting like OpenAI tried to profit off of Scarlett. They tried to profit off of the portrayal she did in the movie Her.
From her statement:
> I received an offer from Sam Altman, who wanted to hire me to voice the current ChatGPT 4.0 system. He told me that he felt that by my voicing the system, I could bridge the gap between tech companies and creatives and help consumers to feel comfortable with the seismic shift concerning humans and Al. He said he felt that my voice would be comforting to people.
So, they wanted to profit off of her voice, as her voice is comforting. She said no, and they did it anyway. Nothing about, "come in and do that song and dance from your old movie."
The only reason that nuclear energy/weapons are possible at all is because there exists isotopes that release more neutrons than they absorb during fission (the chain reaction). Nuclear fission itself wasn't fully understood until 1938.
I think this is the hard part of analyzing systems. If there exists mechanisms outside of your current understanding, there is no way to predict what is and isn't possible
"It always seems impossible until it's done." - Nelson Mandela
Treating features as independent mutations may incrementally improve the software, but this approach can quickly lead to an incoherent codebase that is difficult to modify without introducing regressions. Proper domain modeling is done so we can organize around patterns.
Unlikely to double. The low level API exposes all capabilities, the high level API exposes a subset of those capabilities under a smaller surface. The high level API will not be as large as the low level.
Flight crew have other things to do then count 200-odd squirming people and are flight attendants supposed to memorize unoccupied seats or carry a clipboard up and down the aisles?
A head count is often done with a simple mechanical clicker by the flight attendants right at the plane's doors and/or during the check walk looking for seat belts and unstowed luggage during taxiing. It takes no additional time at all.
Any pilot should have an actual count of the souls on board, it can become a question of life and death in an emergency situation. Would you want to be the one left behind trapped in a burning wreck because "all 200" passengers on the list have been declared clear of it - with the blind passenger taking your spot?
I'm aware this is a made up scenario that probably would never occur even in the event. Take it more as food for thought why knowing the SOB might be more than unnecessary theatre.
I'm really astonished by a lot of responses in this thread, seeing how people that are forced to take off their shoes while waiting in line ay the airport keep on defending willful negligence at seemingly every other point in the process.
Yes, having a full count of people on board seems like a low effort necessary action to be taking especially for emergency scenarios. I assume if the passenger count doesn't match the expected count, especially in the case of a double passenger scan it should flag a more extensive audit.
> Sourcegraph’s maintained fork of Zoekt is pretty cool, but is pretty fearfully niche and would be a big, new infrastructure commitment.
I don't think Zoekt is as scary as this article makes it out to be. I set this up at my current company after getting experience with it at Shopify and its really great.
I wonder if there's a browser extension that checks if the link text is a valid URL, but is a different URL (or just on a different domain?) than the actual link target, and adds some kind of warning for the user if so?
I'm not sure what keywords I'd use to find an extension like that.
This would break every website that wants to track what links you click on by sneakily rewriting the link under your nose. Which, to be fair, is a use case that I'm all for breaking, but it would make Google mad, so it won't happen.
Opus got it correct for me. Seems like there is correct and incorrect responses from the models on this. I think testing 1 question 1 time really isn't worth much for an accurate representation of capability.
Just ran the test and seems to have gotten it correct.
Okay, let's think through this step-by-step:
We know that Sally is a girl and she has 3 brothers.
Each of Sally's brothers has 2 sisters.
Now, who are these sisters? They must be Sally and one other sister, because if Sally's brothers had any other sisters, they would also be Sally's sisters, and the problem doesn't mention that.
So, if each of Sally's brothers has 2 sisters, and one of these sisters is Sally herself, then Sally must have 1 other sister.
Therefore, Sally has 1 sister.
There are a lot of good and bad reasons to adopt a mesh. Some of which might relate to your concerns. The things I like most about them, working in infrastructur:
1. I can have a unified set of metrics for all services, regardless of language/platform or how diligent the team is at instrumenting their apps.
2. I can guarantee zero trust with mTLS, without having to rely on application teams dealing with HTTPs or certificates.
3. I can implement automation around canary releases without much lift from dev teams. Other projects leverage these capabilities and do it for you as well.
4. I can get the equivalent of tcpdump for a pod pretty easily which I can use to help app teams debug issues.
5. I can improve app reliability with automatic retries and timeouts.
Probably some other things as well... That said, it can be a big increase in complexity to your system the pains of which aren't always distributed to the folks getting the benefits.
> 1. I can have a unified set of metrics for all services, regardless of language/platform or how diligent the team is at instrumenting their apps.
But they're only going to be coarse metrics, like what requests/second. You're still going to be needing application-specific metrics.
> 2. I can guarantee zero trust with mTLS, without having to rely on application teams dealing with HTTPs or certificates.
I do like the idea of this feature of service meshes. It is a slog to get teams to do this responsibly. But, like I said: fighting Istio to understand why it was RST-ing a connection, for no apparent reason. Not logging errors is worse. Perhaps the idea is sound, but the implementation leaves one desiring more.
I should mention the same Istio service mesh above is a SPoF in the cluster it runs in, on account of being a single pod. I can't tell if the people who set it up were clueless, or if that's the default. I suspect probably the latter.
> 3., 4., and 5., as well as actually using mTLS in 2.
TBH, these are just benefits I've never been able to realize. I'm stuck slogging through the swamp of service mesh marketing and the people who want to bring the light of their savior the service mesh but without actually getting their hands dirty doing the work of deploying it.
The fact that you need other metrics does not substract from OPs original point. It's still good and much better overall to handle a set of comprehensive metrics at infra level than to orchestrate every app.
About the coarsness, i think it's not really true. Proxies are freaking powerful and they do a lot of stuff at l7, too much in fact (look at envoy, jesus christ). That's one of the reasons why despite the insane complexity of service meshes, they are paramount for observability.
1. I agree with the sibling comment, that's usually not the depth I need for application level info. So the entire skill and infrastructure cost for metrics still exists.
But it's nice to have as basic data for triage.
2. I always wonder whether that's a timing thing vs. NetworkPolicies and encrypted inter-node traffic.
Are the realistic attack scenarios where it's possible to read out intra-cluster traffic but not mess with the cluster, or even read the intra-pod traffic?
3. I've been quite disappointed with how little k8s provides here. I wish it was easier to move traffic off of an old version and only shut the pod down once the last connection was done :/
Maybe I need to look into a service mesh for that?
4. What's the difference to e.g. kubeshark, or just attaching a tcpdump debugcontainer to the pod? Another instance of first to market / potentially nicer ecosystem?
5. I get squeemish with infra-level activities like this.
Yes, technically the http method and some headers should make it obvious whether that's save or might break at-least/at-most once or similar semantics.
But that requires well behaved applications. While the premise here is infra imposing behaviour to allow applications to be looser around these kind of things.
