Susi O'Neill
@susioneill.bsky.social
210 followers 290 following 990 posts
Tech communicator, speaker, author, thereminist, early electronic music nerd. Weekly tech & AI insights: https://rethinkhypecycle.substack.com/
Posts Media Videos Starter Packs
www.mckinsey.com/about-us/new... For cybersecurity folks, AI threats are coming for you so fight fire with fire and get ready to rumble (not quite how McKinsey phrase it though wish they did )
www.mckinsey.com
Small language models can beat large ones (LLMs). Samsung is rocking a teeny model designed for reasoning/ problem solving. It loops back on the solution to provide more reliable outputs with fewer resources. The future for business AIs could be small but beautiful. www.forbes.com/sites/ronsch...
Samsung AI Research Team Builds A Tiny Model With Big Power
Samsung’s AI lab in Montreal new Tiny Recursive Model with only 7M parameters performs as well, if not better in some instances, than systems much larger.
www.forbes.com
www.reddit.com/r/PromptEngi... Possibly these hacks, like asking it to respond as expert with high IQ, may work to get less generic answers from your LLM. The debate also calls bull the idea of 'gaslighting' an AI.
From the PromptEngineering community on Reddit
Explore this post and more from the PromptEngineering community
www.reddit.com
What jobs do people think is acceptable to be replaced by AI and which are morally unacceptable? Good news for caregivers. Search marketers may want to look away.
youtu.be/_zfN9wnPvU0?... This looks like AI slop but isn't, explaining why AI disinformation is shitting in the pot to make AI sourced info less accurate and credible.
AI Slop Is Killing Our Channel
YouTube video by Kurzgesagt – In a Nutshell
youtu.be
As we enter the frenzied holiday season (sorry, too soon), Adobe study shows huge uptake of LLMs in guiding to a perfect gift. And surprisingly, it's driven by desktop users
business.adobe.com/resources/ho...
Disclosing you use AI in your work erodes trust. But disclose you must. So lead with how AI supports your expert endeavours www.nobodycaresaboutethics.com/blog/ai-disc...
www.nobodycaresaboutethics.com
Some handwringing and some opportuities for limitless storytelling (though a worrying lack of railguards - "we'll bring in an AI ethics person after launch" from a storytelling app) - how kids, parents and corporate are using AI to entertain children. www.theguardian.com/technology/n...
‘My son genuinely believed it was real’: Parents are letting little kids play with AI. Are they wrong?
Some believe AI can spark their child’s imagination through personalized stories and generative images. Scientists are wary of its effect on creativity
www.theguardian.com
openai.com/devday Lots of good tings at the OpenAI DevDay - Codex = a self-contained coding and testing tool, AgentKit = no-code agent builder and a SDK to integrate apps inside ChatGPT. Here's hoping this smoother running integration will help those building next-gen apps.
OpenAI DevDay 2025
Explore all the announcements from OpenAI DevDay 2025, including apps in ChatGPT, AgentKit, Sora 2, and more. Access blogs, docs, and resources to help you build with the latest tools.
openai.com
hbr.org/2025/10/what...

This study shows that if an algorithm determines your rate it could find in your favour. It priced up freelance rates and removed gender pay gap. But bias around age and geography persist, with 30-somethings getting the biggest pay boost.
What Happens When AI Sets Wages
Generative AI is no longer just automating text; with AI-based wage-setting, it’s shaping paychecks, livelihoods, and access to opportunity. Left unchecked, these systems risk reinforcing, or even wid...
hbr.org
AI is unsurprisingly popping up in corporate filings for S&P 500 big biz. The problem: most can't explain what it's doing, or how it's benefiting the business. Looks like a prime case of FOMO. futurism.com/future-socie...
Big Businesses Have a Major Problem With AI
An analysis from the Financial Times makes a damning case of mindless AI band-wagoning among the US's top companies.
futurism.com
www.sustainabilitybynumbers.com/p/ai-footpri...
A deep dive on latest research on AI energy use. Search isn't as energy intensive now as previously thought, making up a fraction of daily energy and similar to search. Though data centres is a different environmental problem to consider
What's the carbon footprint of using ChatGPT or Gemini? [August 2025 update]
A new study from Google suggests its Gemini LLM uses around 0.24 Wh per text query. That's the same energy as using a microwave for one second.
www.sustainabilitybynumbers.com