January in AI is your insider briefing on the most important stories from the last month, what actually happened, and why it matters for real work, real decisions, and real systems.This episode covers the push to get AI out of laptops and into the physical world, including Nvidia’s new platform for autonomous vehicles and the broader “physical AI” vision behind it. We also look at what it means for London (and Europe more broadly) as robotaxis move closer to reality on roads that are dense, irregular, and far less car-centric than most US cities.On the healthcare front, ChatGPT and Claude both launched new health offerings. I break down what these tools claim to do, why people will use them, and the two hard questions underneath: data incentives and reliability. Alongside that, we look at Stanford’s SleepFM research and what it suggests about predicting disease risk from a single night of sleep, plus the practical questions that it raises.Then we move to jobs and skills: global AI adoption is rising but uneven; McKinsey is testing “AI-assisted thinking” in graduate hiring; and entry-level roles are getting harder to access for a mix of reasons, not just AI. We also cover the counter-argument: demographic decline as a long-term force that changes the labour market math.Finally: synthetic media ownership (including trademarking likeness), AI-generated music with consent, policing and AI hallucinations, software stocks reacting to AI coding tools, Anthropic publishing Claude’s constitution, warnings about a dangerous phase of AI capability, ChatGPT citing AI-generated “encyclopaedias,” and the arrival of ads inside ChatGPT.If you want monthly sense-making like this, subscribe. And tell me in the comments which story matters most for your work.Read the article: (coming soon)

This is a public episode. If you would like to discuss this with other subscribers or get access to bonus episodes, visit www.thehumansintheloop.ai

ChatGPT ads change the relationship between an AI assistant and the person using it. In this AI Power Move video, Heather Baker breaks down the best-case and worst-case outcomes of ads inside conversational AI - looking at the story from both the optimistic and the pessimistic point of view. From the optimistic view, advertising creates a repeatable revenue stream to fund expensive frontier models and safety work. It could also subsidise access, keeping high-quality AI available to people who can’t afford paid tiers. Heather also explores the argument that a conversational interface might give small businesses a fairer shot than search or social, because user intent and relevance can matter more than pure budget.From the pessimistic view, the story looks familiar. Ad auctions tend to get more expensive over time, and the platform captures the upside. More importantly, once money enters the conversation, recommendations can start to feel suspect: is a suggestion there because it’s the best option, or because someone paid? Heather also digs into the surveillance concern: ads need targeting, and in a chat interface, the targeting signal is what people type and say - including sensitive, high-trust topics.The core question is simple: can an AI assistant be a trusted thinking partner and an ad-driven marketplace at the same time? Let me know where you land in the comments.Links:AI Edit: https://www.theaiedit.ai/ (for all my content on AI concepts explained easily)OpenAI: Our approach to advertising and expanding access: https://openai.com/index/our-approach-to-advertising-and-expanding-access/Can OpenAI be trusted? With great power comes great responsibility: https://www.thehumansintheloop.ai/p/openai-and-anthropic-want-your-medical

This is a public episode. If you would like to discuss this with other subscribers or get access to bonus episodes, visit www.thehumansintheloop.ai

AI hallucinations explained in 2026: what they are, why they happen, how often they show up, and what you can do to reduce the risk.An AI hallucination is when an AI model makes something up: fake facts, fake citations, fake studies, or real things with wrong details. In this video, I break down clear examples (including a made-up “idiom” that ChatGPT confidently explained as real), plus public cases where AI hallucinations created serious consequences: fabricated book lists, lawyers citing court cases that never happened, and high-stakes decisions made on confident-sounding output.You’ll learn why large language models hallucinate in the first place: they aren’t looking up truth from a database (and it's not as simple as saying "it's a feature, not a bug"). They generate text by predicting likely next words, and they’re rewarded for answering fluently, not for being honest about uncertainty. That makes errors feel persuasive, especially when you’re tired, rushed, or already leaning toward a conclusion.We also look at how common AI hallucinations still are, using recent benchmarking, and why “just trust the model” is not a strategy, particularly in legal, medical, financial, or reputational contexts.Finally, I walk through practical risk controls to keep you safe from the consequences of AI hallucinations: treating outputs as drafts, verifying anything consequential, setting human checkpoints and accountability inside teams, and using approaches like retrieval (RAG) to ground responses in sources you control. Hallucinations won’t fully disappear, so you need processes that assume they will happen.Links:AI Edit: https://www.theaiedit.ai/ (for all my content on AI concepts explained easily)AI hallucination benchmark study: https://research.aimultiple.com/ai-hallucination/OpenAI paper (Sep 2025): https://openai.com/index/why-language-models-hallucinate/WTF is an LLM Anyway (video):

Implementing AI in your business (video):

This is a public episode. If you would like to discuss this with other subscribers or get access to bonus episodes, visit www.thehumansintheloop.ai

AI in healthcare is moving from “people asking chatbots questions” to formal health products that connect records, explain test results, and support clinicians. This video looks at what that could mean -both the upside and the risks - and asks a simple question: how much AI do you actually want involved in your medical care (or your child’s)?From the optimistic perspective, AI could reduce administrative burden, speed up care, and help patients prepare better for appointments. It could also support research by spotting patterns earlier across large datasets, and potentially make health guidance more accessible through low-cost, multilingual tools.From the pessimistic perspective, these tools centralize the most intimate data in our lives -diagnoses, lab results, sleep, food, symptoms, and more - inside companies with enormous incentives to monetize. The video explores what privacy promises are worth over time, how partnerships could influence recommendations, and what happens if medical conversations become a new advertising surface.Then there’s the reliability problem: hallucinations still happen, and in health contexts the consequences are real. The video also looks at how training choices shape outcomes—whose worldview is embedded in “health,” and what gets excluded.Would you connect your full medical records to an assistant today—yes or no? And what would it need to do to earn your trust? Share your view in the comments.AI Edit: www.theaiedit.ai

This is a public episode. If you would like to discuss this with other subscribers or get access to bonus episodes, visit www.thehumansintheloop.ai

Humans in the loop is the difference between AI that works safely and AI that causes real damage. This video breaks down what it means, why it matters, and how to implement it.AI systems fail in three fundamental ways. They inherit biases from training data. They hallucinate, making up confident but completely wrong information. And they're sycophantic, prioritising agreeability over accuracy. When these systems run autonomously, the consequences can be costly and dangerous.This video walks through two real-world disasters: Air Canada's chatbot that invented a refund policy and lost in court, and a dealership chatbot that agreed to sell a Chevy Tahoe for $1. Both show what happens when AI acts alone.The solution is designing systems where humans guide, review, and override AI at critical points. You'll learn the ACA framework: Assess, Checkpoints, Accountability, for integrating AI where it adds value while protecting against risk.If you want to understand why AI behaves this way, watch the linked video WTF is an LLM Anyway. It explains hallucinations, bias, and sycophancy in plain English for non-technical people:

Become a top 1% AI user and thinker at the AI Edit: www.theaiedit.aiRead: Anything you say to an AI note taker can and will be used against you: https://www.thehumansintheloop.ai/p/anything-you-say-to-an-ai-notetaker

This is a public episode. If you would like to discuss this with other subscribers or get access to bonus episodes, visit www.thehumansintheloop.ai

A brutally honest conversation about AI sales strategies as Heather Baker sits down with Aleasha Bahr, founder of Black Sheep Sales and author of Black Sheep Sales: How to Win More Sales Without Acting Like a Douchebag. 📘 Get Aleasha’s BRILLIANT book:https://amzn.to/3LoiYBcIn this conversation, Aleasha breaks down why traditional sales tactics are broken, manipulative, and quietly destroying trust. She explains what “douchebag sales” really looks like, how to identify nightmare clients before they drain your business, and why authenticity consistently outperforms pressure tactics.The discussion dives deep into the role of AI in sales, where it actually helps, where it absolutely fails, and why “personalisation at scale” is mostly a lie. Aleasha and Heather explore the idea of the anti-audience, how being controversial can clarify your positioning, and why the best sales strategies focus on fit, boundaries, and long-term relationships rather than chasing yeses at all costs.This interview is essential viewing for founders, sales leaders, and teams who want to improve sales conversion without sacrificing integrity, reputation, or their sanity.👤 Guest: Aleasha BahrFounder, Black Sheep Sales🌐 Website: https://aleashabahr.com/🔗 Explore the AI Edit to become a top 1% AI user and thinker:

https://www.theaiedit.ai/

And check out Heather Baker’s advice on building a sales strategy with AI https://www.thehumansintheloop.ai/p/you-can-sell-more-with-ai

This is a public episode. If you would like to discuss this with other subscribers or get access to bonus episodes, visit www.thehumansintheloop.ai

The last AI briefing of 2025. I run through the biggest stories from December that actually matter: from DeepSeek's free model threatening to disrupt the entire AI pricing model, to Anthropic's $300B IPO prep, Disney's $1B OpenAI deal, and why Wall Street still thinks AI stocks will climb 10% in 2026.

If this helped you, consider hitting the ❤️ to let me know it was worth your time.

I also break down why you should be deeply sceptical of self-reported AI productivity studies (hint: people are terrible at measuring their own productivity), the "slop" economy, and what happens when AI tries to recap TV shows (spoiler: it hallucinates).This is a monthly session. Join me for the January session. Register here.

This is a public episode. If you would like to discuss this with other subscribers or get access to bonus episodes, visit www.thehumansintheloop.ai

In this podcast, I interview Tim Dixon, who is a recruitment leader and entrepreneur with two decades of experience building and scaling high-performing teams across Digital, MarTech and now Generative AI.

He began his career in 2005, rising to Board level within ten years, leading multi-office teams of 70+, managing £26m+ in annual revenue, and securing long-term supply agreements with clients including Sky, BAE Systems, Aviva, AXA, Leonardo and the NHS.

In 2020, Tim founded Beagle, the first and only executive search firm dedicated to GenAI. Beagle has since helped scale global SaaS and creative groups such as Pencil, Jellyfish, OLIVER (Brandtech), BBH and Epsilon (Publicis), redefining how talent is identified and deployed in a rapidly changing industry.

Over his career, he has overseen more than 7,000 hires, consistently developing high performers and pioneering new routes-to-market for both greenfield and established business models. Tim has been focused on leadership roles for the last 10 years and GenAI team building for the last 3 years.

https://beagletalent.com/

This is a public episode. If you would like to discuss this with other subscribers or get access to bonus episodes, visit www.thehumansintheloop.ai

What are the actually useful AI automations for small businesses – and what’s just spam at scale?In this episode of Humans in the Loop, I sit down with fellow South African and automation expert Ferdie Bester, founder of Nightjarr, to unpack what’s really changing in digital marketing with AI, and where the real opportunities are.We get into how large language models like ChatGPT and Gemini are reshaping the buyer journey, why traditional Google Ads and SEO are under pressure, and how smart brands are already adapting. Ferdie shares how his team is using AI to qualify leads, speed up sales workflows and remove manual work – without handing everything over to bots or annoying people with personalised spam at scale.

This is a public episode. If you would like to discuss this with other subscribers or get access to bonus episodes, visit www.thehumansintheloop.ai

What if Silicon Valley’s biggest promises - AI breakthroughs, self-driving cars, billion-dollar unicorns - are built on wishful thinking?

Emmanuel Maggiori is a CTO, author, and serial myth-buster who’s spent years on the inside of the tech industry. In this interview, he dissects the hype and exposes the reality behind the biggest stories in tech.

He explains:

* Why self-driving cars aren’t actually self-driving

* How tech bubbles are fuelled - and who gets burned

* What it really takes to build a unicorn

* The dark side of VC-backed innovation

* Why scepticism is a superpower in today’s tech landscape

Listen if you’re ready to question the hype.

This is a public episode. If you would like to discuss this with other subscribers or get access to bonus episodes, visit www.thehumansintheloop.ai