> I'd also add, maybe controversially, that anyone using "AI" to make some kind of basic prediction (say for personalization, demand planning, medical decisions, really most tabular data) is not going to have a business that lives or dies based on AI.
Unfortunately this rules out the vast majority of business use-cases. ML use-cases on structured data vastly outnumber those of vision/nlp/RL.
Also you seem to be restricting the term "AI" to Neural network-type techniques.
> Also you seem to be restricting the term "AI" to Neural network-type techniques.
I'm curious to know what you are implying with this. Are there commercially relevant startups you know using other "AI"? I'd definitely be interested in a symbolic reasoning startup or some other more obscure AI idea, but I suspect there are vanishingly few so it's unlikely to come up in the OP's job hunt
Not really symbolic reasoning. I'm thinking more along the line of classic ML techniques like Random Forest/GLMs have long been successfully applied to business problems.
The key difference, I think, is that software is user-facing, whereas hardware no longer is. The bigwigs and bean counters who make the hire-versus-outsource decisions can tell that lowest-bidder outsourced software sucks and so don't go for it, whereas the outsourced hardware "works" well enough to ship, so they keep outsourcing it.
Interesting. I wonder if this distinction applies within software as well. For example, systems programmers getting paid less than application programmers because the latter makes software "closer" to the user, and backend programmers getting paid less than frontend programmers for the same reason.
Yes, backend engineers do get paid more in general. I was just wondering why this runs contrary to OP's explanation that "closeness" to user triumphs job difficulty (when it comes to leverage/salaries).
> If the Federal Reserve sticks to its guns and gets rates up, and keeps them there, it will be unsurprising to see "hard" engineering jobs be valuable again, while all the CSS people suddenly can't find two pennies to rub together.
I was under the impression that "hard" engineering isn't as financially attractive to investors due to the long payoff period as compared to web software. In that case, won't high interest rates actually hurt "hard" engineering due to a greater discount being applied to future cashflows?
Precisely; electrical engineering as a field requires enormous amounts of capital. Software development also requires capital, but not nearly as much, and the returns are much higher.
> For instance building some state control engine for rockets etc.
Just out of curiosity: are there companies actually writing mission-critical code in a Haskell? My impression is that those normally tend to get written in C++ and the like.
As they say, technically correct is the best kind of correct but also the most useless.
Parent's point is well received: it's much harder to be a startup when you need to fund laboratories, complex supply pipelines, diverse and deep know-how, interaction with government agencies etc.
One of the more surprising combos at university was Physics and Music. That was the degree, and the numbers were similar to those doing Computational Physics.
Imperial College actually used to offer a BSc Physics and Music Performance course[1] (it was suspended recently). This is notable because dual majors, even between similar/adjacent academic fields, are rare for UK undergraduate degrees.
