How about ship stuff now even if imperfect. Then as capable models come online, ask them to refactor. This allows you to deliver NOW and refactor later when the situation demands.
Satya Nadella reveals why every company may need its own AI model: the model becomes the new company database.
"To me, a model is like the database market."
"A firm should be able to take the tacit knowledge it has and embed it inside weights in a model that they control."
"When somebody asks me how many models should there be, I'll say as many models as firms in the world."
The contrarian part: the value may not sit in one universal frontier model. It sits in each company turning its private operating knowledge into a controlled model.
There are many research avenues which are open which reduces cost dramatically. Smaller task specific/ language specific/ domain specific models, in fact they could even be better. The earlier computers were the size of a building. So prediction based on current state into the unknown future possibilites is wrong. The hardware will be all the more valuable if cheaper ways to run become possible. The hardware gets cornered in a sense.
I wish Rich people did this more often. Not just rich people but rich companies. Not just rich companies but rich governments. But we are a broken society. People should be paying more to OSS for building digital infrastructure.
"Bernard, I have served eleven governments in the past thirty years. If I had believed in all their policies, I would have been passionately committed to keeping out of the Common Market, and passionately committed to going into it. I would have been utterly convinced of the rightness of nationalising steel. And of denationalising it and renationalising it. On capital punishment, I'd have been a fervent retentionist and an ardent abolitionist. I would've been a Keynesian and a Friedmanite, a grammar school preserver and destroyer, a nationalisation freak and a privatisation maniac; but above all, I would have been a stark, staring, raving schizophrenic."
Here's another example, let's say we got the news from Andromeda galaxy that Andromeda Hitler is killing lot of people? What do you expect me to do ? Since space and time are equal, similarly we don't lose sleep over bad events that happened in the past.
I agree with you. Self-help industry has evolved to cover lot of nuanced topics. Depression, anxiety adjacent topics. Once the basics are recognized, the other aspects like what to expect when you grow old. what to expect after retirement etc.
How to care less about what other people think, but in a healthy way
Can we measure preventive medicines or projects which have been designed with care from the beginning? Can we have metrics and measure/compare projects over lifetime production bugs?
This reminds me of perverse Amazon practice of hiring people only to fire them to meet quota. This could have been avoided if they measured the average lifetime of hired individuals. If they were fired the very next firing round, it means the process is broken. Ideally a fixed percentage of freshly hired developers should have moved upward displacing senior slacking developers.
We should create a generalized version of "Gell-Mann Amnesia". This applies not just to fields of study. But also to time and space. We read history as if the person who wrote the history book has perfect knowledge.
I did something different. Instead of describing the image, I described the artist. Made that artist in Ubermensch. Then asked AI to draw the image from his point of view. It worked fabulously.