18 Comments

When I saw that latest OpenAI demo I found the voice extremely irritating. I don't want to have pointless fake conversation with my phone about what I'm wearing or the grin on my face. The fake, over-acting laughing would drive me nuts. I would instantly tell that thing to be quiet.

Expand full comment

Really? I actually found the voice to be right, appropriate, and soothing, depending on the context. It only felt like it was mildly annoying when the person interacting with it wasn't super casual or genuine (I mean, it's pretty hard to be "genuine" when telling a robot that you're talking about "it" because it has no sense of agency or consciousness or self); actually, the fact that Sky reacts that way shows how sophisticated its sentiment analysis is, because it somehow implies that it recognizes that it is not a real entity, and the only way to respond to the information that the launch is about "it" is to be slightly "fake" (I mean, think about it, there is no other way to make the interaction fluid without making it slightly awkward for most people).

I thought Scarlett's voice in "Her" was actually fake and overacting (trying to be sexy but failing, at least to me). This is a crucial consideration in copyright infringement claims. If you can show that Sky's voice has more substance and variety than Scarlett's in "Her," then the work is not really a derivative or copy of Scarlett's voice (assuming people would argue that), merely on the basis that it is much "bigger" and "better" and "more creative" than Scarlett's voice. I think if it ever comes to that, the case is automatically won by OpenAI.

However, I do not think the lawsuit will get to that point, and the executioners of OpenAI's legal counsel will get this case under wraps pretty easily. I mean, the case against Scarlett is really easy to win, she pretty much has no case there and it's all unrelated circumstantial evidence. I mean, you cannot possibly think that any judge is going to take the arguments "oh, they contacted me and Sam tweeted 'her' before the launch and the Internet and OpenAI haters are with me on this and think the voices are super similar" as evidence that OpenAI did anything wrong there. I have gone to YouTube and compared the voices of "Her" and GPT-4o in several instances and they are not similar at all (which I bet would translate into an even bigger variation in a granular audio profile of the voice frequencies).

To be honest, I am with OpenAI here. I do not understand the hate against them. How long have we been waiting for personalized assistants that feel real and respond quickly?

Expand full comment
author

Daniel it's okay if you miss the original Sky. She was wonderful in her own way.

Expand full comment

I see what you are doing there :) Yup I miss the original Sky? :p

Expand full comment

The computer on Star Trek TNG is what I think we need. Not “Her” or anything like “Her” even. And I think most of what I saw did not interact in a way that seems useful when you really want to get a job done. Again, in Star Trek, see how the computer actually helps them solve problems. I’m not impressed by an AI that can tell me what I’m wearing. I already know what I’m wearing.

Expand full comment
author

Sounds like Anthropic to me. Enterprise do you know Claude?

Expand full comment
May 22Liked by Michael Spencer

Everyone is entitled to their own opinion naturally so I understand your apprehension however, I personally like technology becoming more human than robotic so if that means that the AI assistants become more natural to speak with then I'm all for it. You mentioned "Her" so taking the examples of the technology demonstrated in that movie, I much prefer the technology from "Her" than say "Minority Report" with technology being subtly there in our world such that it does not interfere with us interacting with the world around us instead of it being right in our face and plastered on the world around us, the cities, buildings and so on. More for "Her" than say "Minority Report" and "Bladerunner". The subtle balance and marriage between technology and nature is better for us if we must travel down that path.

Expand full comment
author

Unfortunately if things were only that simple. Curiously U.S. exceptionalism in creating an AI Monopoly startup actually limits innovation in the US and enables China to build a better ecosystem of innovation around generative AI.

If openai is the poster child of Monopoly capitalism in the AI era and it's eating most of the revenue in its space, it's not actually a healthy thing for the future of generative AI.

That's not what free market capitalism is or is supposed to be. So OpenAI can you use the social hacks and have more funding than the others, but ultimately it's not a great sign for how the US is taking the category down a pretty manipulated path.

Expand full comment

Definitely, the competition from others such as Google and those abroad will be interesting to see. For the moment, its great to see the advancements but scary about the ethical challenges that will come up if not regulated.

Expand full comment
May 21Liked by Michael Spencer

If the script to that teaser video about ChatGPT4o was written by, say, ChatGPT-4, then Hollywood has nothing to fear from Sora. I was so bored watching it in a group of 20 people meeting about AI for SMEs that I didn’t see the punchline - my mind was somewhere else more interesting (like Orson Wells b/w film The Third Man).

Expand full comment
author

I know right, my mind drifts to various Black Mirror episodes and all the dystopian scenarios where all of this corrupts society and increases Technological loneliness and reliance on predatory data hungry technology.

Paul, I must be getting older, I'm struggling to see the Technological optimism angle in all of this.

Expand full comment
May 21Liked by Michael Spencer

Mostly I see small beer, not big beer - Open-AI’s top table of management is behaving in far too dysfunctional a way to be building a sustainable business (because that’s what it is).

Expand full comment
author

Yes and the more funds they raise, the more pressure they will be under until some kind of mean reversion and diminishing returns is reached if it hasn't already. The conduct of Sam Altman in particular seems concerning. When the cash cows run out, it's going to be an epic collapse.

Expand full comment

Perhaps OpenAI is now primarily funded by the US government and that's why a lot of talent is leaving the company and why they can now more or less make everything free to use.

Expand full comment

Their ties with the US military and pentagon are truly pretty fascinating. In the era of seeing China as a competitive threat anything goes.

I believe in the end the EU will be a bit tougher on Microsoft and openai then the rest of the world.

Another narrative is that it's free because ChatGPT is bleeding users and time on site. 18 months later it's not a novel product and for most people not that useful.

Regardless what we think this is an extremely revenue generating growth company. It's so successful that consolidation with other foundation model builders might accelerate to the point of a duopoly by 2027.

It, generative AI, becomes another layer on top of the cloud and digital advertising that is fairly important for some of the most valuable companies in the world.

Expand full comment

You alright Mike? This article reads like ChatGPT 2 wrote it with spellcheck and proofreading disabled.

Expand full comment
author

No I'm not alright Elyas, I take this conduct by OpenAI to heart. It's Black Mirror personified.

Expand full comment