Nonsense on stilts. Extrapolating $3500 in your first week of media-driven traffic into a $15k/mo run rate is ridiculous. Calling it "AI-powered" is a whopper too.
Extrapolation aside, making decisions automatically based off of incoming data points _is_ AI.
If the interview is to be believed, almost anything you consider "AI-powered" is within one degree of what is happening here. At least on a conceptual level.
AI doesn't need to be magic, it just needs to be making decisions :)
It's not just his definition. Many AI courses will teach you pathfinding algorithms as examples of AI, along with optimisation algorithms, genetic algorithms and neutral nets.
There is a valuable distinction in hard AI vs soft AI (hard AI being that whole thinking, emotional maschines thing that we are not really getting closer to, and soft AI being the things that actually generate money because we know how to do them)
I think what he's doing would be more acurately described as Machine Learning, rather than AI. His algorithms are producing better logo proposals based on the data it is collecting from other users, which fits a common definition of ML [1]
[1] A computer program is said to learn from experience E with respect to some class of tasks T and performance measure P, if its performance at tasks in T, as measured by P, improves with experience E (Tom M. Mitchell, 1997)
Machine learning is a better term because it's more specific (and less controversial). But under most definitions of AI, ML is a subfield of AI, so calling it AI may be suboptimal, but not wrong.
Personally I'm a fan of the AI definition "things humans can do and computers can't do yet", but I recognize that that definition isn't terribly useful.
The distinction you are drawing is between Strong and Weak AI. Weak AI is any sort of useful application of computation. Strong AI is solving the AGI (Artificial General Intelligence) problem.
anything that doesn't work but theoretically could is AI, when it starts working it stops being called AI. most recent examples are convnets, deep learning, etc. nobody calls them AI anymore.
He is doing a weighted decision. But that is not AI.
AI is more complex and based on Neural Nets, Petri Nets, or alike.
But he is just applying the same logic like every webshop, or advertisement network is doing: you bought X, other people who bought X also liked to buy Y. There is a long way to make AI.
If you pull out old AI books, they're all about search spaces and agent systems. Basically an abstraction of graph search.
Your definition requires calling the first couple generations of AI research "not AI".
And I would most definitely call recommendation engines Artifical Intelligence (what is Google?)
To paraphrase an old quote "AI stops being AI when people start understanding how it works".
Is it not a replacement for a guy standing in a room and trying to keep the temperature at the right spot?
I understand that there's "current-generation AI" which is miles beyond this. But closed-loop systems that try to make decisions based off of loop seems very much in the domain of intelligence in the practical sense.
I'm being a bit pedantic here, but the GGP in this thread saying that OP is not AI feels like shifting the goalposts way beyond what we would have said even 5 years ago.
Control systems are not intelligent. It helps to know how they work, and when you do you realize they work on very simple principles.
Intelligence is about learning and applying knowledge, AI is either simulating or synthesizing these behaviors outside of nature. It's not just taking the place of a function that something intelligent does.
