a drug that does help or something that helps. So huge development. I mean, wow, what a step forward in our ability to give people the right medicine. So there, an AI model that can predict whether an expensive drug will work. That's big stuff. This should have been like a world headline. I mean it's so important.

So how'd it go? Well, not the way you hope. Not the way you hope. So he developed this model and then the implications are dramatic. Instead of wasting a whole year taking the pill to find out it doesn't work, AI predicts in advance the efficacy for each patient. So how did the company react? This is still in the tweet from Dr. Eli David. They asked him to delete the project and never mention it to anyone. The AI model would strip them of huge profits generated by millions who take the drug and find out it's not effective for them. Better let them suffer and pay the company instead.

What do you even say about this? Well the first question you should ask is, is it true? Right, I just told you that fake news is a big problem. You already knew that. Is it even true? Is it? Well there is one problem with the story, right? A little bit too on the nose, isn't it? Because isn't it exactly what you expect, right? So it's a little too on the nose.

What's the second part of the story that should raise a red flag? Second part of the story that raises a red flag. Let's see if anybody gets it. What would raise a red flag in this story that he wouldn't tell anyone? Well he did tell someone but he quit to do it allegedly. AI model that works. Here, that is a wonderfully cynical but accurate comment. Somebody said that the strangest part of the story that should tip you off is fake is that somebody made an AI model that works. That's not a bad comment, right there. Yeah, I'm going to endorse that one. Yeah, somebody made an AI model that works. Really? I mean I'm sure it's happened but you know right there you're gonna — I don't know, it's kind of hard.

Are you missing — there you go, you're almost there. Somebody said rogue doctor. You know whenever there's the rogue doctor you know you have to — you should put the odds of them being right at very low. But it's not a rogue doctor, it's an AI person so a little bit different. Didn't name the pharma company. Well probably did name the pharma company to the writer. Anonymous. Okay, okay. We don't know if he's anonymous by the way. Yeah, we don't know that. The tweet doesn't name him but I don't know if he will always be anonymous. But that is exactly the right thing you should be looking at. Thank you. Thank you. We got the right answer here from Bear Wires.

Disgruntled employee. Do you believe disgruntled employees ever? Do you know how often you should take what a disgruntled employee says as just fact? You know, there's no context left out. Boom, that's a fact. How about never? Never would be a good time to take a disgruntled employee's opinion. How about never? About not one time ever. Now it doesn't mean they're always wrong. I'm just saying that in terms of credibility it's kind of zero, right? Again, doesn't mean they're wrong. It's a right or wrong is a separate question from whether it's the kind of thing you should believe when you hear it. I'm saying it's the kind of thing you shouldn't believe when you hear it. It has that nature: disgruntled employee, not named company, an AI model that works. A little too on the nose right now.

If it's real — and by the way I don't want to cast any aspersions on Dr. Eli David. I have no reason to think that he would have the story wrong or anything like that. So I'm sure that the person he talked to is a real person that he really talked to who really worked for a big pharma. I'm sure he did the basic work. So I think the basic facts are probably true. He doesn't work for a big media company that maybe is bought off in some way as far as I know. I mean I guess you could never know, huh?

But suppose that's true. And how about this: let's take this and extend it. Do you think it would ever be possible to write an AI, create an AI that could spot fake news for you? What do you think? Could you ever write an AI program that would help you spot fake news fairly accurately? Maybe not every time. Yes, the answer is absolutely. It just wouldn't work every time. How do I know that an AI could do that? Because we just did it. We just did it right here. We just came up with several objective standards that an AI could just remind you of. It's an anonymous source. It's a disgruntled employee. It's an unusual claim that AI works. Yeah, I'm exaggerating a little bit there but you know if this AI worked as well as reported it would be quite a surprise to me that somebody got that done.

And then there's the question of whether it's too on the nose. Could an AI identify a story that was too on the nose? Not directly but it could do it by a poll of humans. It could ask humans, "Hey, does this story match pretty much your suspicions or does it not match the suspicions you already had?" And if the AI finds out that a whole bunch of people go, "Oh yeah, that's exactly what I expected," then the AI can say, "Oh, I have now demonstrated it's a little bit too on the nose. It's a little bit too exactly what you thought was going to happen, isn't it?"

It could also look at the sources. The AI could say, "Okay, Fox News says it's true or false and CNN says it's true or false." And unless both of them agree on the facts, it's probably fake news. Just an objective standard. If the left and the right don't agree on the basic facts, don't assume it's true b