Nope. Sorry. I'm trying to stick to doing work I can be paid for these days, and I don't know enough about tech, algorithms, and law to even address the issue. I will leave that up to those much smarter than I am.
Yes, this is the very very verbose me back for moment.
So, I have taken some time out to skim the surface of AI. And I do mean skim the surface. I just wanted to get a better idea of what it is, what it's doing, and where it's heading. I don't suggest anyone else should do it, but I have a natural curiosity. Also, you don't get much from just skimming.
So what have I seen?
Well, firstly I saw a working application that does the following: You take a video of yourself, or use a video from someone else, as a base. In the example, it's a video of a guy walking through what appears to be a park next to some office buildings. The guy is talking to the camera. Having pulled this up in the application, you can then use your webcam/camera to record say, gestures (raising a hand, pointing at something, putting your arms above your head, etc.) This is the run through the app, and when replayed, the character in the video perfectly replicates your new gestures.
This app goes one step farther too. You can also give it new speech via text. You provide the speech/text into the system and it will lip-sync your model to the new words, which it will use. So now you have the original character, walking through the park, but he now raises his arms, and speaks - using a natural sounding voice - the words you give him. But it doesn't quite end there either, because since this is essentially turning your text to speech, the language doesn't matter. So if you give it Spanish, the character will speak Spanish, German, French, whatever.
What does this show/prove? Well, it shows that any video you now see of celebrities saying or doing things, need to be questioned. Don't take them for granted. You cannot trust videos posted on the Internet at face value. The model had examples of using Obama as a model, for example. They also used Elon Musk to advertise a fake investment scheme.
Second thing I saw was a simpler model that can do the same but essentially only the lip-sync bit. This is most useful if you simply want to use an interview segment from a celebrity to send a different message (change the speech/words used).
Next up, I saw a empathetic voice system demo. Essentially, we all know the robotic voices that call us on the phone asking for our passwords, or used on Help Lines, right? Well, this system can interpret the tone and timbre of your voice. In other words, it knows if you're angry, sad, depressed, inquisitive, being humorous, etc. It then models itself on human behavior to respond. So for example, if you're angry, it will use a calming voice tone. If you're sad, it shows a sympathetic voice tone, etc.
Then I learned that the BBC has, in at least one case, fired a female presenter from one of its shows because it has decided to use an AI generated voice instead. In fact, the voice-over industry (voice actors) are apparently extremely worried - their days are numbered! Turns out the BBC isn't alone, and other actors/voices have recently lost their jobs. Why? Well, the obvious, of course. It's cheaper, easier to control.
What does this show? Voices you hear on TV/videos may or not have been spoken by the actual person. In fact, it may not be a person at all.
I moved on to seeing a demo of an image creation system (and by system, I mean software). In this one, you upload an image say, of your dog (which was the example used). Let's say it's a picture of your dog standing in a field. The system then takes that dog, and is able to create new images based on it. So you get pictures of your dog jumping, running, drinking water, sitting up, laying down - all generated by the software, all photo-realistic.
A secondary to this was another system that not only models the image (as above), but also takes into account things like light source, weather conditions, etc.
What does this tell us? It tells us that the pictures we see may, or may not, be true. And no, it doesn't have to include Photoshop. In fact, if I were Adobe and owned Photoshop, I'd be mightily worried around about now.
I need to do more research on Neural Networks Scaling Laws and Entities. Though I don't think there's any interest in knowing a whole lot about AI on these forums.
I also found a survey where more than 40% of managers said AI would be taking over some jobs in their workplace in 2024. The jobs affected are low paid tasks rather than management. So yes, there will be a huge shift due to AI coming. It'll be justified as these things are always justified - it's about cost cutting, and what good capitalist doesn't want to do that?
This means, if you getting into the workforce now, you're going to need to skill-up. I guess it's always been that way, which is why we have education programs.
Finally I'll say - this all comes back down to my oft-made complaint about discernment. Our protection is our eyes, our curiosity, and our intelligence. We should not jump on something just because it appears on a Social Media account you subscribe too, or even if it goes viral on like-minded Social Media accounts. If you truly want to know the truth, rather than jumping on a bandwagon, you have to cast your net wider. Because increasingly, you're going to see and hear things on the net that are fabricated for effect. Not misunderstood, but actually 100% fake.
Anyway, I thank you for the inspiration to at least do a tiny tiny bit of looking into this stuff, it's fascinating.
