There Is No Bubble

4 points by mattwiese 1 year ago | 5 comments

PaulHoule 1 year ago |

God. "Imagine what the insurance industry can do with cheap compute to personalize premiums for better margins. Or what the unloved healthcare industry can do to maintain patient outcomes while minimizing labor costs."

I mean, really? Insurance industries could probably optimize their underwriting by a few more percent than they do, but I can't see big data being all that transformative. Same in health care. A cure for Alzheimer's disease or better drugs for schizophrenia, or affordable GLP-1 drugs for everyone could change the picture -- big data won't. It won't solve the problem of getting all the crazy folks on the street to get a depot injection once a month, for instance.

mattwiese 1 year ago | |

"Big data" has been and continues to be transformative. It's part of why we are in this current AI cycle. After each fad passes, what remains forms the foundation for the future. This sums it up:

> The economy functions in cycles just like any other ecosystem - we didn't stop building Internet companies after the dotcom bubble popped. If anything, it eliminated uncompetitive species (companies) while opening up ecological niches (markets) for those with the fitness capable of exploiting them.

PaulHoule 1 year ago | | |

There are a few winners and a lot of losers.

No doubt Facebook's personalized advertising throws off so much money they can afford to burn a lot of it.

On the other hand, I think you're doing 23AndMe a favor if you tell them to delete your data because it's close to worthless.

bediger4000 1 year ago | |

LLMs will not get us any of the good things you mention. They might get us slightly faster claim rejections, all worded suspiciously similarly.

PaulHoule 1 year ago | | |

That’s the point. We heard all this hyperbolic talk about big data in the 2010s that didn’t come true —- GE boasted how much data was collected by their turbines for instance and, yeah, you get pretty much the same data on revolution 1,000,001 as you do on 1,000,000. AI and ML will play a role in new drug discovery but you are still going to have do a lot of work in the lab to develop them and still spend a lot on clinical trials to validate them.