I came across this article. The title says it all:
I have just four thoughts
As always for journalism, there is no empirical verification if AI researchers smoke more than people of the same socioeconomic status. Nor if there is a rise in such smoking. This is the classic anecdote turns into headline which is the preferred method of journalism, a field that lacks quantitative skills. But accepting this story at face value…
In a world of biotech, where so many people don't believe what they say, it's refreshing to meet someone who's willing to put skin in the game. Famously, Michael Debakey declined the very aortic surgery he invented. And I'm willing to bet that if Elizabeth Holmes needed an important blood test, she wouldn't put it in a theranos machine. It's worth acknowledging that the AI researchers who are panicking about the end of the world, or prophesizing cures for all diseases— are true believers. It's not just something they're telling investors. Side note: if you think the world's going to end in 10 years or you think cancer is going to be cured in 10 years, both are perfectly legitimate reasons to start smoking. Either way you escape the consequences.
As much as I'm bullish about AI disrupting medicine, I'll stick to not smoking. The odds ratio for smoking and lung cancer is tremendous. Everyday, we debate minuscule odds ratios. A recent paper said pilots are more likely to get radiation-induced cancers. People told me they will consider limiting their flying. Lol. Ok. Consider these are minuscule risks, that probably aren't even true. After all, cancer doesn't divide itself into radiation-induced and non-radiation-induced cancers. Only researchers do that after the fact if they don't find an elevated all cancer risk. And the odds ratios observed are piddly. In contrast, smoking and lung cancer are tremendously associated. Smoking and cardiovascular events are decently associated. Next, I'm not sure AI is going to cure lung cancer in time. It's possible that ill-advised Congressional action will soon regulate the space. There'll be a federal agency of AI that ensures real innovation is stifled and a massive amount of bureaucracy slows progress. Even without governmental sabotage, AI might not figure it out in time. When it comes to coding, playing chess, doing mathematics, AI doesn't need anything more than it already has. But when it comes to physics, biology, and chemistry, AI may need the results of experiments that do not yet exist to fully develop its hypotheses in thinking. Anthropic is building a laboratory whose pi is AI. It's hard to know how long before AI has these insights. I'm not going to gamble.
Psychiatric conditions like anxiety, depression and schizophrenia have always associated with cigarette use. There are some psychiatrists, old and nearing the ends of their careers, who believe that the pharmacologic effects of cigarettes have benefited their schizophrenic patients. It would be interesting to note if these traits/ tendencies are more common among AI researchers, and if they derive psychological benefit from smoking. The line between genius and madness has always been thin.
The whole point of technologic advance is it allows you to enjoy life. Instead of washing your clothes in the river, you can toss it in a machine and go for a walk in the forest. Instead of losing 4 days traveling 100 mi, you can drive there in an hour and enjoy more time there.
AI will do the same thing. Instead of combing through your emails to pull out all your 1099s, AI will just prepare your taxes. Instead of doing menial PowerPoint slides for work, AI can do that.
Ironically, as of now, the people I know who use AI the most have increased their workload the most. Maybe they crave a cigarette. As for me, I don't. There are plenty of other ways to enjoy life.
Trusting AI to cure cancer so you can smoke today, is like trusting the government to ensure social security is solvent so you can retire tomorrow. I wouldn't take my chances.
If you like this, substack become a subscriber.
Our university pays for an upgraded version of Claude…I’ve tried to implement it in my work and have found it overhyped and underwhelming. There was an initial “gee whiz” period, after that has subsided it has not proven very useful for me. I do use it for completing annoying administrative paperwork that I don’t care about. But anything that I do care about, I do myself. After all, if I didn’t bother to write it, why should you bother to read it.
it is refreshing to see people putting their own skin in the game... unlike the pools (obama, gates, Gore, sanders) who fly around in private jets, on the waterfront estates crowing climate change nonsense