> Its not disguised. Corporate blogs exist overtly to promote the company and its work.
It is. That makes the "research" heavily biased. If xAI did the same thing, with Elon Musk screaming about that it is "AGI", you would not believe them at all.
Given that the work is not independent, such articles of this "research" can easily be manipulated or the results being massaged to promote the company positively.
But when others outside of the company try out the work or reproduce it, they get different results. So of course we continue to hear unverified research especially in AI when the frontier labs do not release their architecture, weights at all.
So in this case with labs raised with VC-funded cash, the incentives are clear and I would not straight up believe results from the first party source unless multiple sources outside of the company have verified it.
You or some other interested person could go do that experiment and publish the results. It shouldn't be hard to figure out what hardware exactly they were using and get a copy, and the prompt also doesn't have to be exactly what they used, just similar enough in spirit. See just how similar/different the outcome is.
Using their own time and money to disprove corporate propaganda. By the time you’ve disproven it they’ve already released 10 new claims. “Why are you still talking about that old news, nobody cares.”
It’s like the firehose of lies but done by corps.
In this case at least it isn't that hard. Opus is available to pretty much everyone, and anyone of sufficient means (I'm guessing at least 95% of active HN) can also easily afford the hardware.
And obviously this isn't something that's being iterated on rapid-fire; there have been 2 relevant publications roughly a year apart. Absolutely no firehose of anything here. As such there should be no problem for someone with enough interest to attempt to disprove the claims, and hopefully share the results regardless of their findings.
You’re writing with the assumption that this is “research” in the first place. This is advertising first, “research” second.
> If xAI did the same thing, with Elon Musk screaming about that it is "AGI", you would not believe them at all.
I’m not saying it is trustworthy or that I believe them, I am saying the advertising isn’t even a little bit disguised when it is communicated directly from what is overtly a promotional channel for the company involved.
It's like calling the “9 out of 10 dentist prefer...” claim in a TV commercial “disguised advertising” and then coming back with arguments about how it isn't trustworthy reaearch wheb it is pointed out that TV commercials are openly ads. Yeah, its not trustworthy, but the fact that it is corporate promotional material and not a neutral third-party report is not at all concealed.
It is overt advertising communicated through a channel whose sole and open purpose is advertising for the company whose products it advertises.