Good AI

…implementations. Now the hubbub [in lieu of a. using in lieu of improperly as so often is the case, b. ‘hype’] is settling down a bit, though some laggards are still at it instead of keeping mum and just doing it in a proper way.

The proper way is that the real talk is of now.

  1. On the one hand, there’s an enormous lack of use of the tools that are available now – not widely available, but they’re there, if one would take a look, and see their building ‘tool’ status. No, I don’t mean the glitzy stuff with fancy screens. I mean the ‘advanced’ stuff, that captures the forefront of what was available in theory functional descriptions and ready-made demonstrators of … yes, twenty-five years ago.
    Those have been transformed, by scraping off the (then) practical drawbacks of lack of training data, of compute power, and comparative economic advantages v.v. human labour. The latter, having come down not in cost [your mileage may vary when comparing purchasing power ..!!] but in intelligence. Yes. Read my many posts on ‘management’, cooperative efforts, et al. The fact (sic) is that now, the marginal out-performance of ‘systems’, gains interest.
  2. On the other, there’s the realisation that as with any hype, the proof is in the long, arduous, classic systems implementation drudgery of building and rolling out actual ‘systems’ – not only the software/hw as that is to be somewhere in a cloud [even if private, localised, i.e., a data centre of yours] or you’re doing IT wrong, but also including the assembly line operators duly called ‘managers’ (them again!) that think they still control stuff but are allowed to tighten a screw here and there and certainly not do more as they shouldn’t interfere with the system since they only can/will do wrong then – the system being the assembly line of information flows, of bits and bytes that should be untouched as that’s where the money is in not in the mere auxiliaries called humans. For lack of a better pejorative word.
    This then includes Change Management of the organisation kind, sometimes jokingly called ‘process‘ change. Which is harder to ‘get’ than understanding (true; sic) AI technology; the latter being much beyond mere theoretical physics Sheldon, let alone rocket science. But still quite easy when one puts one’s mind to it. Unless you’re not even believing in yourself for probably a very good reason.
  3. On the third hand [don’t even bother to smallmindcorrect me], there’s more serious stuff to do around the system engineering. Take it as a normal business investment project, like here. Ah, now there‘s the clue to this post. But without the above intro, it wouldn’t been as much fun, wouldit’vebeen..?

So, finally again:

[Now go contemplate, e.g., where this is. Any half-decent AI implementation will render ‘Convento Corpus Christi, V.N. de Gaia’ in seconds; what about you ..?]

Leave a Reply