It seems odd to phrase things as "hype" and say Apple doesn't target engineers as their target market.
The machine literally has a component in it called a neural engine which Apple advertises as:
> In fact, with a powerful 8‑core GPU, machine learning accelerators, and the Neural Engine, the entire M1 chip is designed to excel at machine learning.
So asking "how well does that really perform" seems like precisely the right question, and to me it'd seem very clear Apple wants a piece of that market.
The Neural Engine is intended solely or almost solely for inference, not training. For instance, in a post from last November [1], Apple mentioned their tensorflow_macos fork can use the "the GPU in both M1- and Intel-powered Macs for dramatically faster training performance", but didn't mention the Neural Engine. (Incidentally, I think that is the same fork used for the benchmark here.)
Anyone who is serious about ML (and by serious I mean like actually training fairly big models with lots of data for a project or as part of their job) is going to use cloud resources or at least have a decent gpu(s) at home. At no point will they ever look at getting a laptop to do ML one. Not only are you going to pay extra for less performance, but even with CPUs (which are still important for ML since you need to not only do the matrix math on GPUS but also prepare the datasets and set up the GPUs), you are not going to reach even close to the power that desktop chips can put out in any laptop with any manufacturer.
As for general targeting of engineers the history of MBP pretty shows that they don't really care. Things like virtual escape key on the touchbar, shitty keyboards with sticking keys, hardware designed to not be reparable, and more recently, releasing the M1 models with a spotty backwards compatibility of software out the door is really not something that would be done if you were targeting tech minded people.
Does any of the software mentioned in the article even utilize the so-called neural engine? If it's just generic software running on CPU/integrated GPU, especially if under Rosetta (no mention about that in the article,) the results represent only the current software support situation, and not much else.
The machine literally has a component in it called a neural engine which Apple advertises as:
> In fact, with a powerful 8‑core GPU, machine learning accelerators, and the Neural Engine, the entire M1 chip is designed to excel at machine learning.
(https://www.apple.com/mac/m1/)
So asking "how well does that really perform" seems like precisely the right question, and to me it'd seem very clear Apple wants a piece of that market.