Conversation

Notices

1. Embed this notice
   Kris (isotopp@infosec.exchange)'s status on Friday, 14-Nov-2025 18:13:05 JST Kris

   A company I know of is now in the 3rd wave of "introduction of AI" and it is entirely unironically a fundamental success that has raised productivity and quality by an order of magnitude.

   Here is how that worked:

   The first wave of AI in this company was an unmitigated disaster. It turned out that the data was unclean, of bad quality, and to top it off, also unlabeled. Also, many processed were unclear in what actually was a deliverable, what the quality metrics were and what exactly had to be done with the data. Processes were similarly nebulous.

   AI was completely unable to do any work, much less successful work, because it was largely undefined what the work was, and what data was present, or what success as.

   After the chaso of the 1st wave and the reorg and redefinition of all data pipelines in the 2nd wave (with exploding AI cost), now the 3rd wave is a huge success. Also, AI cost went down dramatically and processing speed went up, because AI could be removed from many data transformations – it turned out that now many things could be done algorithmically in a traditional way, as it was clear what actually had to be done.

   AI has magically transformed this organization into a much better and more successful org.

   • GreenSkyOverMe (Monika) repeated this.
   • Embed this notice
     Vaui (vaui@mastodon.berlin)'s status on Friday, 14-Nov-2025 18:17:29 JST Vaui

     in reply to

     @isotopp
     Whenever you hear “A.I.”, replace it with the term “Automation”. The actual algorithm of automation then does not matter anymore, but rarely any automation works on bad input.

     Reminds me of my time working on mathematical optimization: Six of eight weeks was data preparation and only one week algorithm develpment and one week its implementation.

     Nice side affect: When using Automation instead of A.I., you automatically have to clarify WHAT is actually automated.

   • Embed this notice
     Kris (isotopp@infosec.exchange)'s status on Friday, 14-Nov-2025 18:20:06 JST Kris

     in reply to

     • Lars Stitz

     @stitzl You don't say! 🙂

     But AI is fine, if it is successfully enabling a company to access it's own meta and actively motivates the org to reflect about what it does and how it does that.

     That's quite the point I was trying to make.

   • Embed this notice
     Lars Stitz (stitzl@mastodon.social)'s status on Friday, 14-Nov-2025 18:20:07 JST Lars Stitz

     in reply to

     @isotopp If I read that, I get the impression that it was not AI by itself that transformed the company into a more successful org, but it was the fact that the had to analyse, redesign and refactor their processes and dataflows to reach a goal they deemed worthy. Because they had an intrinsic motivation, they persisted, where otherwise they would have probably given up half-way.

     GreenSkyOverMe (Monika) repeated this.
   • Embed this notice
     GreenSkyOverMe (Monika) (greenskyoverme@ohai.social)'s status on Friday, 14-Nov-2025 18:20:32 JST GreenSkyOverMe (Monika)

     in reply to

     • Ronnie Soak

     @Chaos_99 Yes, that is what Kris is saying.

   • Embed this notice
     Ronnie Soak (chaos_99@fosstodon.org)'s status on Friday, 14-Nov-2025 18:20:35 JST Ronnie Soak

     in reply to

     @isotopp So the 3rd wave of AI is ... less AI? Because doing it badly with statistics showed that it's possible to do it better with deterministic algorithms?
     Am I reading this correctly?