I liked the last part, about saying that something is AI generated when it isn't. Has anyone created a tool that can change a picture or a text in such a way that an "AI detection tool" thinks it is AI generated? Just add a sixth finger, add the word "delve" or something like that...
Well actually, you can take this one level higher: create an image with AI, ensure it is so good that it is not detected, and then use the aforementioned tool to change it so it is detected as AI. Then release it. People will assume it is AI, until a digital sleuth will find out it has been changed. And then everyone will think it is real, while it isn't!
And your second point, oof, this all seems like it'll be so messy. Where there can be layers too it. And also, at what point is something real or faked? Instagram filters? Sorta fake? I assume that stuff will all use AI at some point too, so will there be generally acceptable AI-based adjustments, like recoloring and stuff, and not acceptable ones, like changing the entire background? I don't know, it all gets so weird.
I noticed recently that Instagram has added a "created with AI" toggle, which is effectively an AI "honour system", to "help" or inform people when content was developed via AI tools. But honestly, it's not going to prevent anyone from jumping the turnstiles for entry. AI detection tools are going to be more of a thing I'd reckon in the coming months (or weeks, since this stuff seems to move faster than a microwave).
Everything you're writing definitely reflects my observations and feelings. People have a choice. Sail into the chaos of the AI storm and risk likely death, or watch from the outside and risk death. In the storm you focus on the sailing, adapting to the chaotic ocean but keeping an eye on the horizon, looking for the lighthouses in the distance that will guide you to success. It's the balance of those two things that has always been the difficulty and I don't think that has changed. The risk/reward extremes have changed greatly though and that's unlikely to become less extreme any time soon.
Good luck everyone and may the odds be ever in your favour!
Basically? And I think that’s more or less the advice I’d give people: That whatever you decide to do now probably won’t work, but you’ll be able to make a much better decision about what to do after that (or after that) if you’re getting tossed around the ocean today. So it doesn’t make that much sense to stress about finding the perfect thing now; instead, just put yourself in a position to learn, and make a more informed bet later, when you’re smarter and when the water is more predictable.
Now the question is will AI allow investors and VCs to duck out of deals that weren’t a good idea? My AI clone signed that not me? I imagine that could get kind of ridiculous…
Nice to hear my suspicion confirmed around the current Silicon Valley environment. I am also fascinated to learn if the rules are the same, slightly different, or in some way completely new….
My guess is that it'd be really hard to get out of something like a VC deal, but I'm sure someone will sue someone for something *like* that. Like the startup used AI-calculated metrics and put a disclaimer that they were doing it and they might be wrong, and the metrics were wrong, so they sue the company for fraud. And a hundred other versions of stories kinda like that.
Interesting! I had not seen the air Canada thing. Ya - I had a boss back in the day that would have people sign stuff and then say later they werent authorized. 😔🫣
Had not imagined a false fake... Since the PW is now "MAGA2024!", we probably have some fascisty statements in our future, optionally retracted if too much for Americans, and written off as poor password hygiene.
But also, you probably sink like a stone if you try to swim too, because the water is weird. (But sink enough, and you’ll learn to swim before everyone else? That’s the hope, I guess.)
I liked the last part, about saying that something is AI generated when it isn't. Has anyone created a tool that can change a picture or a text in such a way that an "AI detection tool" thinks it is AI generated? Just add a sixth finger, add the word "delve" or something like that...
Well actually, you can take this one level higher: create an image with AI, ensure it is so good that it is not detected, and then use the aforementioned tool to change it so it is detected as AI. Then release it. People will assume it is AI, until a digital sleuth will find out it has been changed. And then everyone will think it is real, while it isn't!
The analog version: https://x.com/bristowbailey/status/1625165718340640769
And your second point, oof, this all seems like it'll be so messy. Where there can be layers too it. And also, at what point is something real or faked? Instagram filters? Sorta fake? I assume that stuff will all use AI at some point too, so will there be generally acceptable AI-based adjustments, like recoloring and stuff, and not acceptable ones, like changing the entire background? I don't know, it all gets so weird.
I noticed recently that Instagram has added a "created with AI" toggle, which is effectively an AI "honour system", to "help" or inform people when content was developed via AI tools. But honestly, it's not going to prevent anyone from jumping the turnstiles for entry. AI detection tools are going to be more of a thing I'd reckon in the coming months (or weeks, since this stuff seems to move faster than a microwave).
That's pretty funny, and reminds me this bit: https://www.youtube.com/watch?v=XaqOejIaFgM&t=697s
Everything you're writing definitely reflects my observations and feelings. People have a choice. Sail into the chaos of the AI storm and risk likely death, or watch from the outside and risk death. In the storm you focus on the sailing, adapting to the chaotic ocean but keeping an eye on the horizon, looking for the lighthouses in the distance that will guide you to success. It's the balance of those two things that has always been the difficulty and I don't think that has changed. The risk/reward extremes have changed greatly though and that's unlikely to become less extreme any time soon.
Good luck everyone and may the odds be ever in your favour!
Basically? And I think that’s more or less the advice I’d give people: That whatever you decide to do now probably won’t work, but you’ll be able to make a much better decision about what to do after that (or after that) if you’re getting tossed around the ocean today. So it doesn’t make that much sense to stress about finding the perfect thing now; instead, just put yourself in a position to learn, and make a more informed bet later, when you’re smarter and when the water is more predictable.
who wrote the ‘end of software’ google doc link? love it, had same thought, but thinking software now ‘is’ the AI you custom train and grow
It's from here: https://x.com/cpaik/status/1796633683908005988
Now the question is will AI allow investors and VCs to duck out of deals that weren’t a good idea? My AI clone signed that not me? I imagine that could get kind of ridiculous…
Nice to hear my suspicion confirmed around the current Silicon Valley environment. I am also fascinated to learn if the rules are the same, slightly different, or in some way completely new….
That's basically the story of that Air Canada thing (though for a much smaller deal), and they were held responsible for it: https://www.cbsnews.com/news/aircanada-chatbot-discount-customer/
My guess is that it'd be really hard to get out of something like a VC deal, but I'm sure someone will sue someone for something *like* that. Like the startup used AI-calculated metrics and put a disclaimer that they were doing it and they might be wrong, and the metrics were wrong, so they sue the company for fraud. And a hundred other versions of stories kinda like that.
Interesting! I had not seen the air Canada thing. Ya - I had a boss back in the day that would have people sign stuff and then say later they werent authorized. 😔🫣
Oh no 😯 I hope everyone's ok out there 🙏
Had not imagined a false fake... Since the PW is now "MAGA2024!", we probably have some fascisty statements in our future, optionally retracted if too much for Americans, and written off as poor password hygiene.
https://www.bbc.com/news/technology-55337192
Easy password to guess, because it definitely wasn't gonna be his kids' names.
Lol this was fun. It’s like you are writing a new episode of Silicon Valley sitcom every time I read one of your substacks.
Sadly, that show is a more accurate representation of Silicon Valley than anything I could ever write....
better start swimming... https://youtu.be/90WD_ats6eE?si=tgP6WZQ_7WLT__zM
But also, you probably sink like a stone if you try to swim too, because the water is weird. (But sink enough, and you’ll learn to swim before everyone else? That’s the hope, I guess.)