e516 — Model Behavior

Games At Work dot Biz

Photo by insung yoon on Unsplash Published 2 June 2025 e516 with Michael and Michael – AI prompts, browsers, model collapse & automation along with a teeny tiny pico Mac nano, and a whole lot more. While Andy is away, Michael and Michael start off with an article from Ars Technica that explains the system prompts for Anthropic’s Claude 4 models.  This leads into a discussion on prompt engineering, and how solutions like Ollama allow users to download LLMs and create their own prompts.  After a quick sidebar on AI browsers like Opera, the team takes a look at Sky, an AI automation app.  This app shows a great deal of promise as a desktop AI assistant, and will be very interesting to try out once it is generally available.  Then, the team turns to a story on AI model collapse.  And next up is a blog post about consenting to updated terms and conditions. Round things off for this episode, Michael and Michael enjoy a teeny tiny Mac classic – the Pico Mac Nano – a new take on 3D monitors and a local North Carolina story about Pokemon card game competitions. What would you want to run on a Pico Mac Nano?  Sky’s the limit!  Have your bots 🤖 drop our bots 🤖 a line at @gamesatwork_biz@mastodon.social (our home for now) and let us know! These show notes were lovingly hand crafted by a real human, and not by a bot.  All rights reserved.  That’s our story and we’re sticking to it. Selected Links AI Ars Technica article: Hidden AI instructions reveal how Anthropic controls Claude 4 Wikipedia article: Prompt Engineering OLlama Games at Work e429: Promptly Engineering The Verge article: Opera’s new AI browser promises to write code while you sleep Opera AI Browser MacStories article: From the Creators of Shortcuts, Sky Extends AI Integration and Automation to Your Entire Mac Sky.app Phrase of the moment: "model collapse" https://www.theregister.com/2025/05/27/opinion_column_ai_model_collapse/ — Mike Elgan (@MikeElgan@mastodon.social) 2025-05-29T14:44:13.236Z The Register article: Some signs of AI model collapse begin to reveal themselves "Technology should only ever do exactly what we have explicitly given it our consent to do" This blog post 👉🏻 https://www.anildash.com//2025/05/27/2025-05-27-internet-of-consent/ by @anildash is just superb! 👏🏻 👏🏻 👏🏻 And the last sentence, a massively poignant call-to-action that won't leave people indifferent, specially, those who can't understand #ConsentMatters #Respect #FairUse #SocialWeb #OwnTheWeb — Luis Suarez (@elsua@mastodon.social) 2025-05-27T20:26:57.215Z Anil Dash blog post: The Internet of Consent Hardware TechRadar article: Someone just built the world’s smallest working Mac – and at this price, I desperately want one Pablo Picasso Wired Article: 3D Is Back. This Time, You Can Ditch the Glasses Pokemon The Assembly article: More Than a Card Game Web 11.0 mashup junkie, and co-founder / co-host of the GamesAtWork.biz podcast. My views are my own. Michael Martine

00:00
27:06
00:00
.
Help
·
Privacy
·
Terms
​
  • avatar
    Apple Podcasts
  • avatar
    Spotify
  • avatar
    Overcast
  • avatar
    Castbox
  • avatar
    Pocket Casts
  • avatar
    Podcast Addict
  • avatar
    iHeartRadio
  • avatar
    Podbean
  • avatar
    Audible
  • avatar
    YouTube Music
  • avatar
    AntennaPod
  • avatar
    Podverse
  • avatar
    Podcast Republic
  • avatar
    Player FM
  • avatar
    Fountain
  • avatar
    Podcast Guru
  • avatar
    Pandora
  • avatar
    iVoox
  • avatar
    Downcast
  • avatar
    Deezer
  • avatar
    Podcast App
  • avatar
    Castamatic
  • avatar
    gPodder
  • avatar
    TuneIn Radio
  • avatar
    RSSRadio
  • avatar
    Castro
  • avatar
    LN Beats
  • avatar
    CurioCaster
  • avatar
    Goodpods
  • avatar
    Anytime Podcast Player
  • avatar
    TrueFans
  • avatar
    Podurama
  • avatar
    Moon FM
  • avatar
    Global Player
  • avatar
    Breez
  • avatar
    podStation
Help
·
Privacy
·
Terms
e516 — Model Behavior