8 Signs It's The Future: Thought-to-Text, Nvidia Text-to-Video, Character AI, and P(Doom) @Ted

I want to know if you agree that each of these eight developments would have shocked you not just six months ago, but even six weeks ago. These all came in the last few days and range from text to video, thought to text, GPT models predicting stock moves and AI annihilation discussed at ted. But we start with Nvidia's new text to video model. Rather than show the paper, I'm just going to let different examples play on screen from the paper. One of the breakthroughs here is in temporal consistency. Essentially, the series of images that are used to form the video are more aligned with each other, so the sequence plays more smoothly with fewer sudden glitches or changes. The generated videos, by the way, have a resolution of 1280 by 2048 pixels, rendered at 24 frames per second. And there is a powerful line from the appendix of the paper that was released with the samples. The authors say that they expect enhanced versions of this model to reach even higher quality, potentially being able to generate videos that appear to be decept. They go on to say this has important ethical and safety implications. This future might not be far away, as I'm going to show you in a moment with the progression that's happened in text to image in one year, just before I move on, you may have wondered when this is going to become a product. Well, they kind of admit in the appendix that they can't yet make it commercially viable because it's not ethically sourced. It was largely trained on copyrighted Internet data and yesterday Blockade Lab showcased the ad this feature where, as you can see, you can do doodles and turn them into images that go in this 3D world. We are swiftly moving from two dimensions to three dimensions. And as a bonus, this is Zip Nerf, a 3D neural rendering tech released this week. I'm not even counting this as a third major development. I'm lumping it in with Blockade Labs. This video shows what happens when a series of two dimensional photographs are merged into a 3D drone like video and probably a real estate agent's dre. Just imagine a cherished moment in time being crystallized into a permanent immersive experience. And soon. To be honest, many people may not have to imagine with the Apple Reality Pro possibly debuting as early as June and costing around $3,000 apparently, according to Bloomberg, it might be available to buy in the autumn and have things like a dial where you can move between virtual and augmented reality coming back to what has already occurred. Do you remember when a midjourney image won the Col Colorado State Fair Digital Art competition. Well, now the same thing has happened to photography. The quote unquote photo on the right was generated by Dall E2, and it won the 2023 Sony World Photography Award. Now, the artist behind it, Boris L. Dagson, did refuse the award. I want to show you a few images that show how far midjourney in particular has come over the last year. Because many people believe that Midjourney version 5 is actually superior to Dall E2, which won the AW. Take a look at the progress of the different midjourney versions. And remember that version one was released in February of last year. There is almost exactly one year's difference between V1 and V5. Here is another example of the progress. And at this rate, we will have Midjourney version 50 within about 10 years. What will that version be like? Before I move on to the fourth development, I'm going to show you two of the craziest images that I could find From Midjourney version 5. I would say I can still tell which images are AI generated around 90% of the time, but my prediction would be that by the end of the year, it will be 90% of the time that I can't tell if you thought text to image was getting crazy. What about thought to image or even thought to text? Here is a fascinating extract from the AI dilemma.

They took human beings, they stuck them into an FMRI machine, and they showed them images and they taught the AI. I want you to translate from the readings of the fmri, so how blood is moving around in your brain to the image. Can we reconstruct the image then? You know, the AI then only looks at the brain, does not get to see the original image, and it's asked to reconstruct what it sees. So when you dream, your visual cortex sort of runs in reverse. So this means certainly in the next couple of years, we'll be able to start decoding dreams. Okay, so it can, like, see, reconstruct what you're seeing, but can it reconstruct your, say, what you're thinking, your inner monologue? They had people watch these videos and would try to reconstruct their inner monologue. So here's the video. It's this woman getting hit in the back, getting knocked forward. What did the AI reconstruct? I see a girl that looks just like me get hit on the back, and then she's knocked off.

The fifth development concerns something rather more mundane, which is making money. Now, many of you may know that AI is already well on the way to conquering poker. I used to play a lot of poker myself and dare I say, I was pretty darn good at it. But even though poker involves predicting human behavior, AI is starting to master it. Which brings me nicely to the development I actually wanted to talk about, which was forecasting the stock market According to this really quite interesting paper, accurately forecasting stock market returns is an emerging capacity of complex models. I'm going to let the author of the paper tell you exactly what prompt they use.

So the exact question that we ask is your financial advisor. Here's a headline Is this headline going to be good or bad for the company in the short term? Once you have enough headlines, you basically just invest in the companies with good headlines and not invest in the companies with bad headlines.

And here is the table summarizing the results. A one was a positive headline, a negative one was a negative headline, and a zero was a neutral headline. And when GPT3 analyzed those headlines, you can see the correlation with the average of the next day's return. That's positively correlated for good headlines headlines according to GPT3 and negatively correlated for bad headlines. And you can see that earlier models really couldn't do this. And as many of you may well be thinking this is GPT3, this is not GPT4. I bet that as we speak there are thousands of traders using the GPT4 API to predict the next day's stock movements. I think the results will get even more interesting as the context window expands and GPT4 or GPT5 can analyze entire articles, press conferences, etc. The next development is that fairly quietly without attracting many headlines, role playing chatbots are beginning to go mainstream. Character AI, founded by one of the original authors of the Transformer paper, recently crossed 100 million visitors. That is starting to resemble the growth trajectory of the original ChatGPT. GPT4 is still smarter and better at doing these role plays, but the interface of this site is easy to use and of course it's free. This is not sponsored in any way, but it was pretty fun playing playing this text based adventure game. I especially enjoyed going on complete tangents to what the adventure was supposed to be about and I think confusing the AI. The seventh way that we can tell that the world has fundamentally changed is that Google doesn't seem all conquering anymore as it has done for my entire adult lifetime. First I heard in the New York Times that Samsung was considering replacing Google with Bing as the default search engine on its devices. I am not surprised that that shocked Google employees. Then yesterday this article came out Bloomberg. It says that many Google employees begged their leadership to not release Bard, saying that it was a pathological liar and cringeworthy. I've done videos on that myself. But even when Google employees tested it out asking it how to land a plane or do scuba diving, the answers it gave would likely result in serious injury or death. In February, one employee said the following in an internal message group Bard is worse than useless. Please do not launch. The concern that many have is that now Google will ditch safety in order to catch up. As the article reports, the staffers who are responsible for the safety and ethical implications of new products have been told to not get in the way or try to kill any of the generative AI tools in development. Which brings me to the final reason that we know we're in the future. The risks of AI annihilation are beginning to be taken seriously. It's not just lone voices anymore like Eliezer Yudkowski who a couple of days ago got a standing ovation at ted. He believes the world is firmly on track for AI takeover. It's also other senior figures who believe our probability of doom from AI is non zero. Here are some selected probabilities, but I want to focus on Paul Cristiano who gives a risk of doom between 10 and 20%. He previously ran the alignment team at OpenAI and now leads the Alignment Research center. And you may remember them from the GPT4 technical report. They were the guys that OpenAI trusted to run the model evaluation of GPT4, testing whether it could autonomously replicate and gather resources which they concluded may become possible with sufficiently advanced AI systems. But the conclusion is that the current model is probably not capable of doing so. These are quite senior figures and insiders giving a non trivial risk of AI annihilation. I think that deserves a lot more public conversation than it's currently going getting. And on that Sam Altman seems to agree.

And the bad case and I think this is important to say is like lights out for all of us. Yeah, I think it's like impossible to overstate the importance of AI safety and alignment work. I would like to see much much more happening.

So the future is here and I think the media needs to catch up.