Actually... (because all salvos in pedantry pissing matches must begin with 'actually')
1. "Literal actual psychology research" is not very compelling. It's not like it's actual, peer-reviewed physics research. Psychology (and social sciences in general) have a p-value of threshold of 5%. And, on top of that, a staggering amount of published psychology experiments
cannot be replicated. And for each paper you find claiming that humans are good at detecting lies, another can be found claiming the opposite.
2. That link was to an abstract, not to the paper, so I don't know how large of effect they are claiming to have found. Is relying on instinct rather than looking for tells a
detectable difference, or a significant/useful difference? Sure, I agree that humans evolved to have a subconscious ability to detect lies,
and an innate ability to tell lies. The two largely cancel out.
3. Moreover, I would expect they designed their experiment to minimize the emotional investment of the participants, e.g. asking them to detect lies about something they don't care about. But it's precisely when we do care about the outcome that detecting lies matters, and that investment will sway how we interpret the evidence. C.f. 'confirmation bias'.
4. Finally, the conclusion of that study...that instinct is more effective than training...is the exact opposite of what people in this thread (and others) claim: that you can learn how to detect lies. That it's a skill that improves over time.
I've seen following (non-professionally) this question for years, mostly in the context of law enforcement, and how the confidence police have in their own ability to detect lies leads to forced confessions. Which is why the profession is (too slowly) moving away from bogus training to detect lies through "tells" and toward the more reliable techniques I described above.
But...to try to be self-aware...it could be that my emotional investment in the issue may distort my own perceptions on this question.