Once again, we're in that unfortunate realm, where no proper rigorous metric exists to determine "intelligence". The few measures purporting to be thus are usually crude, in some cases dependent upon cultural assumptions, and have at best a chequered history.

An interesting insight into the issues involved was provided by the early days of research into artificial intelligence, back in the 1960, when Gelernter wrote a computer program whose purpose was to devise mathematical theorems, starting from a small base set of axioms. Eyebrows were raised when, having moved into Euclidean geometry, the program provided a proof of the equality of base angles of an isosceles triangle, that differed from the standard proof taught in textbooks. Gelernter's program caused quite some excitement at the time when it provided this proof, until it was discovered that the same proof had been devised 23 centuries or so earlier by Pappus of Alexandria. Gelernter's program eventually ran out of ideas, so to speak, and the project became a footnote in AI research, one documented by Douglas Hofstadter in Gödel, Escher, Bach.

Of course, Gödel's Incompleteness Proof should have warned the researchers against prospect of the Gelernter program being able to devise anything truly new and wonderful, because that proof constitutes a "barrier of ignorance" which even world class mathematicians cannot cross. That proof, in effect, states that there exist theorems of elementary number theory (and by extension, any other formal axiomatic system with the same expressive power) that are true, but which cannot be proven by mechanical inference from the axioms of the system. This is, in effect, the pure mathematician's version of the speed of light barrier in General Relativity.

A parallel lesson to that arising from the Gelernter program, has yet to be learned in some quarters amid the hype about "machine learning", which in some quarters has become little more than a synonym for pattern matching in big data sets, a process that is frequently brought crashing to a halt by the issue of spurious correlations, an issue that any competent scientist, particularly one with a background in mathematical statistics, will home in on with laser guided precision and subject to ruthless dismantling. Even an elementary student of statistics learns to be wary of coincidences over time being treated mistakenly as correlations, and more subtle, but potentially even more misleading and even dangerous spurious correlations, are causing alarm bells to ring in the requisite academic departments.

Unfortunately, the dangers of mistakenly treating mere coincidence as correlation, a plague that affects supernaturalism to a woeful extent, are being ignored by the quick buck brigade, who are jumping on "machine learning" as the next big booster rocket supposedly propelling them into the world of collecting mansions and yachts. Except that oops, this won't happen in 99.9% of cases, because of failure to apply rigour to the matter of eliminating spurious correlations. Worse still, the unscrupulous may end up propelling themselves into billionaire fantasy land, on the back of practices that are not merely lacking in rigour, but unethical or downright criminal in their implementation, with all the fallout arising therefrom that has been a feature of "neoliberalism".

But I digress. The point should, by now, have been driven home, that decades of research into what intelligence actually is, has thus far turned up a lot of information on what it isn't, but with few clues in the positive direction. Any assertions on the part of those claiming to possess this holy grail, should therefore be treated with a large amount of suspicion.

My own view, and an admittedly speculative one at that, is that a defining feature of intelligence is the ability to manipulate concepts. Mathematicians with a grounding in category theory are already seeing where I'm heading with this particular piece of speculation, namely that increasing levels of abstraction will ultimately separate genuinely intelligent thinking from mere pattern matching, but I admit this idea is speculative, and probably won't bear fruit even if if becomes the focus of a well-funded research programme. However, if someone wants to splurge a fortune on this, and in doing so, actually makes positive progress, I for one will be delighted.

Indeed, that's a central question that has yet to be answered properly even in the academic community, as far as I'm aware - namely, is there any aspect to intelligence that is distinct from pattern matching? Answer that one, and I suspect suitable progress will follow.

I feel like it also wouldbt be far off at all to say intelligence is the ability to understand. Understanding is a very wide range word, and it exist on many levels.

@logicFTW "In my humble opinion:

A good marker of intelligence is the ability to separate fact from fiction with the available information we have."

I would label this as discernment.