AGI is Bananas
A fun conversation this week on TWiT's Intelligent Machines show.
We kick off talking through Google's decidedly smarmy Pixel event. Where yes, the bizarre nature of the production overshadowed anything that was announced.
We then shifted gears to the fate of Chrome. Will Perplexity be able to buy it? (No.) What about OpenAI? (No.) Will the judge force Google to do something with it? Maybe, in particular if he's worried they're already using it to favor Gemini to give Google an advantage in AI in some way – but that world is already so competitive...
Maybe Google should just present a platinum brick to Donald Trump and be done with it. Or let the government take a 10% stake in Chrome going forward.
Speaking of Apple... are they behind in AI? I mean, yes. But does that matter? That's still not entirely clear. Certainly not clear enough to make a, say, $30B+ acquisition. Then again, that's maybe the only thing that can truly instill a mentality shift inside of the company to make sure they can operate in the Age of AI. Unless, of course, the system rejects the organ – as seems to happen with Apple acquisitions for obvious reasons...
Speaking of AI... before I'm willing to say we'll get to AGI one day, it would be helpful to know what that – or any of the other terminology – actually means. It also probably would have been helpful to define the terms before the famous/infamous "clause" was baked into the Microsoft/OpenAI agreement – something which they're still fighting over to this day. And certainly all the endless talk about "AGI" hasn't helped OpenAI in the GPT-5 roll-out.
With my investor hat on, I try to think back to what the launch of ChatGPT was like and how there was quite a bit of trepidation around the general space because we were just coming off of the last crypto boom and bust which burned a lot of people – not just SBF. And yes, we had just been told that VR would work this time and that the metaverse was inevitable – to the point where Facebook renamed themselves! As fun as DALL-E was, was a chatbot really going to be the future of AI? That wasn't so obvious until GPT-3 and then GPT-4 came about...
Is it a bubble? In some ways, sure. It always is. But I wouldn't expect the fall-out to be that bad. This time. This time is different.™
In terms of AI usage, I reveal my new obsession with getting ChatGPT to write me a summary that we can then discuss after every episode of television that I watch. It really is like a personalize post-show hangout podcast, in a funny way. And they're really good!
Lastly, I give the verbal argument for why AI won't replace writing – at least certain types of writing – where the input and what you get out of the process is just as important as the output.