Russel - totally agree. I used it today to set out options for a key business decision. Ran the slide deck and ran the deep dive audio. Both processing in parallel. Just watching the slide deck with audio (1 voice) and then later - listening to the 2 person deep dive audio - gave me a “third person” perspective and made it easier to make the decision. Sometimes a step back and hearing others - really works. It’s quite a product isn’t it.

@Stan Eyler - have a look at this model GAISI - it’s been used in the Uk but adaptable. The principles are more insightful than a blanket “AI will take our jobs”. “ The paper —“How Exposed Are UK Jobs to Generative AI? Developing and Applying a Novel Task‑Based Index”—is indeed real. It was published on July 30, 2025 on arXiv . ✅ What GAISI Measures - GAISI (Generative AI Susceptibility Index) scores jobs on a 0–1 scale, representing the fraction of job tasks where an LLM can reduce task time by ≥ 25% beyond current tools, weighted by task importance as reported in the Skills & Employment Survey (SES)  . - - It leverages 44 tasks across 25 UK SOC occupation groups, with LLMs assigning probabilistic exposure ratings (E1 through E3), and E2/E3 discounted by × 0.5 to reflect latency/latent exposure  . 📊 Key Numbers & Distributions - As of 2023–24, 95% of UK jobs show at least some exposure, but only 13% exceed GAISI > 0.5—meaning very high exposure is rare  . - The typical (modal) job has exposure around 30–40%. - Overall, about 25% of tasks are rated directly susceptible (E1), another ~24% latent (E2), and negligible E3 (< 1.5%)  . - This results in an average GAISI score of roughly 0.26, i.e. ~26% of tasks can be accelerated by chatbot alone; additional ~27% require software integration  . 📈 Dynamics & Labour Market Signals - This rise in exposure between 2017 and 2023/24 stems mainly from occupational shifts toward knowledge‑heavy roles, not from changes in the tasks themselves  . - Wage premium for high‐GAISI roles fell from +4.9% (2017) to +4.4% (2023/24)  . - High-exposure job vacancies dropped ~6.5% after ChatGPT’s release compared to trend predictions  . 🧮 Mathematical Assessment 1. Index Construction & Validity - GAISI uses model‑rated probabilities combined with worker‑reported task importance weights. Multirun reliability (ICC) is high—they ran multiple classification runs with Gemini 1.5 Pro showing stable distributions (E0 ≈ 50.6%, E1 ≈ 24.5%, E2 ≈ 23.8%)  . - Validity: GAISI predicts reported AI use in SES better than other indices—one standard deviation increase raises probability of AI use by 12 percentage points, while other measures become statistically insignificant when included alongside GAISI  .

@Stan Eyler Ha Ha - Stan I've watched EVERY single episode of all Star Trek's and of course know DATA well. There's a reason why Musk is trying to colonise Mars, whilst putting low cost robots and self driving cars on earth (hint: he's watched Battlestar Galactica too much IMHO). 😃

I’ve unpacked it with Claude… This is a fascinating snapshot of where the general public stands with AI adoption. Let me unpack the underlying dynamics and opportunities: The Three Barriers Reveal Different Problems “Not enough time” is rarely about actual time scarcity. It’s about perceived ROI. People make time for things that deliver clear, immediate value. This signals a failure of demonstration - most AI content shows party tricks (generate an image! write a poem!) rather than “here’s how I saved 10 hours this week.” “Don’t know how to create leverage” is the most honest and revealing response. People see AI as a novelty, not a multiplier. They’re stuck in a feature mindset (“ChatGPT can write emails”) rather than a system mindset (“AI can handle my entire customer onboarding workflow”). “It’s moving too fast” is actually good news disguised as anxiety. It means people are paying attention but feel paralyzed by FOMO. They’re waiting for the “right moment” to jump in, not realizing that moment was six months ago. The Real Opportunity There’s a massive gap between: • AI capabilities (which are genuinely transformative) • Public understanding (which is superficial and anxiety-laden) This creates multiple opportunities: 1. The “AI Sherpa” Business Model People don’t need another course on prompt engineering. They need someone to audit their actual work, identify the 3-5 highest-leverage AI applications for their specific situation, and implement them. Think: consultative, done-with-you services rather than one-to-many courses. 2. Vertical-Specific AI Plays Instead of “learn AI,” go narrow: “AI for real estate agents,” “AI for physical therapists,” “AI for small law firms.” Show them exactly what to do Monday morning. The riches are in the niches, and most people want a paint-by-numbers solution. 3. The “Time Arbitrage” Angle Reframe the entire conversation: “You don’t have time NOT to learn this.” Create a 90-day challenge showing before/after time audits. Make the cost of inaction visceral and real.