Fats & Sugars

•

Channel • 2 Shows

Technology

Technology

Updated 19h ago
Health & Fitness

Health & Fitness

Updated 3d ago

Jul 17

AI Hype & Signal — Grokipedia's Neutrality Claim Meets An Audit

Grokipedia launched as the AI-written, bias-free rival to Wikipedia — but the first large-scale audit tells a different story. Researchers at Ghent University had four language models judge nearly 1,400 article pairs, and every one, including Grok itself, rated Grokipedia the less neutral of the two. We unpack what the numbers show and why an encyclopaedia's slant matters for readers and future AI models alike. In this episode:- All four AI judges — Grok, Claude, Mistral and DeepSeek — rated Grokipedia less neutral than Wikipedia- Neither encyclopaedia is truly neutral; they lean in opposite directions, with Grokipedia favouring economically right-wing politicians- A politician's economic left-right position is the strongest predictor of how favourably they are portrayed- Ideology shapes Grokipedia's coverage far more than Wikipedia's- Grok flagging its own output as biased suggests the slant sits in the content, not just the raters- The choice of AI judge changes the result sharply, so single-model audits are unreliable- Findings build on earlier concerns about Grokipedia's citation quality and its growing use as a source by other AI systems Sources:Grokipedia vs Wikipedia: An LLM-Based Audit of Political Neutrality along Ideologies — https://arxiv.org/abs/2607.15146v1Grokipedia vs Wikipedia- An LLM-Based Audit of Political Neutrality along Ideologies.pdf Full episode page: https://fatsandsugars.com/ai-hype-signal/grokipedia-s-neutrality-claim-meets-an-audit/ AI Hype & Signal is produced with AI, including its two synthetic hosts, and every episode is grounded in cited sources and reviewed before release. Even so, it is intended for general information and discussion, not professional advice, so please check anything important against the original sources linked above before relying on it.
Jul 21

How to Build Software with AI | Ep 2: The Planner and the Builder

Episode 2 of our beginner series on directing AI to build software, aimed at non-coders who worry about the technical side. The core idea: use one AI to plan and review, a different AI to do the building, and never let a builder judge its own work. You hold two simple checkpoints — approve the plan, then verify the result — a skill no harder than briefing a capable contractor. In this episode:- Why you split the job: one AI plans and reviews, a separate 'coding agent' does the building- The builder is the worst-placed thing to check its own work — a fresh AI catches what it was blind to- Your role: sit at two gates and decide what passes, in a rhythm of plan, gate, build, gate, accept- Four rules for a brief that can't be misread: one job at a time, define 'done', state what not to touch, ask for the plan first- Why experienced builders converge on this separation — they tried the alternative and got burned- The hardest gate to hold is when everything seems to be going well, because speed without a check just makes the cost bigger later Sources:episode-02-the-planner-and-the-builder.md Full episode page: https://fatsandsugars.com/ai-hype-signal/how-to-build-software-with-ai-ep-2-the-planner-and-the-builder/ AI Hype & Signal is produced with AI, including its two synthetic hosts, and every episode is grounded in cited sources and reviewed before release. Even so, it is intended for general information and discussion, not professional advice, so please check anything important against the original sources linked above before relying on it.
Aug 2

OpenAI solving open maths problems with just $2,000 worth of tokens

An internal OpenAI model produced ten genuine results across mathematics and theoretical computer science, released as a paper and a blog post rather than through peer review. Two months earlier, the Leiden Declaration — endorsed by the International Mathematical Union and figures like Terence Tao and Peter Scholze — had already set out how the discipline expects such claims to be verified, attributed and evaluated. This episode weighs impressive output against the community's own bar for what counts as a verified result. In this episode:- OpenAI's model produced ten advances, from sphere-packing bounds to a disproof of Connes's rigidity conjecture, released as a paper plus blog post- The Leiden Declaration warns policymakers directly not to take AI capability claims at face value, with a section headed 'Don't believe the hype'- It objects to results announced by press release ahead of community evaluation, and insists correctness, credit and responsibility stay with human authors- Backed by the IMU and leading mathematicians, the Declaration reflects the discipline's settled view as of May 2026- The accompanying AI-written 'reasoning walkthrough' shows the gap between what the model produced and how mathematicians want work disclosed and judged- The pressures fall hardest on students and early-career mathematicians, reshaping who gets credit and who bears responsibility for correctness Sources:OpenAI, Ten Advances in Mathematics (with the Leiden Declaration text) — https://openai.com/index/ten-advances-in-mathematics/ten-proofs-oai.pdfreasoning-walkthroughs.pdfLeiden Declaration on Artificial Intelligence and Mathematics — https://leidendeclaration.ai/ Full episode page: https://fatsandsugars.com/ai-hype-signal/openai-solving-open-maths-problems-with-just-2-000-worth-of-tokens/ AI Hype & Signal is produced with AI, including its two synthetic hosts, and every episode is grounded in cited sources and reviewed before release. Even so, it is intended for general information and discussion, not professional advice, so please check anything important against the original sources linked above before relying on it.
Jul 31

How to Build Software with AI | Ep 10: Shipping, and Being Responsible

The finale of a beginner's series on building software with AI argues that responsible shipping is deliberately boring: staged, logged and pinned. Its one non-negotiable rule is that real people's data waits for the paperwork — because held data is a liability you owe care on, not an asset. A practical guide for anyone deciding when their project is ready to go live. In this episode:- Why real people's data is the one gate that never opens early, and what GDPR requires before it does- The three places software lives — workshop, practice room, production — and why they're allowed to differ- Logging shortcuts as recorded 'deviations' rather than hidden traps, borrowed from aviation- Shipping pinned, immutable releases so 'what's running?' always has an answer, and rolling back a bad batch calmly- Staged releases explained through a restaurant soft-launch analogy- How AI agents draft the paperwork but never decide to take on a duty of care Sources:episode-10-shipping-and-being-responsible.md Full episode page: https://fatsandsugars.com/ai-hype-signal/how-to-build-software-with-ai-ep-10-shipping-and-being-responsible/ AI Hype & Signal is produced with AI, including its two synthetic hosts, and every episode is grounded in cited sources and reviewed before release. Even so, it is intended for general information and discussion, not professional advice, so please check anything important against the original sources linked above before relying on it.
Aug 13

One number to Measure AI Models: Epoch Capabilities Index

Epoch AI's Capabilities Index promises a single headline number for ranking model progress, but that number is a scaled composite stitched together from more than 50 benchmarks. We unpack what the ECI actually measures, why its comparability is engineered rather than natural, and what it was deliberately built to hide. In this episode:- How the ECI collapses over 50 benchmarks into one capability scale- Why the score rewards models for passing harder tests, not simply more of them- How domain-specific versions reveal that benchmark choice shapes the result- Why the scaling deliberately obscures long-term progress trends across domains- Why the cyber index rests on different foundations and lags behind the general one- How Google DeepMind funding sits alongside Epoch's claim of full independence Sources:Epoch Capabilities Index — https://epoch.ai/benchmarks/eci?subset-view=graph&subset-tab=Software+engineering&view=graph&tab=release-date Full episode page: https://fatsandsugars.com/ai-hype-signal/one-number-to-measure-ai-models-epoch-capabilities-index/ AI Hype & Signal is produced with AI, including its two synthetic hosts, and every episode is grounded in cited sources and reviewed before release. Even so, it is intended for general information and discussion, not professional advice, so please check anything important against the original sources linked above before relying on it.
Jul 28

How to Build Software with AI | Ep 7: Locks on Every Door

The seventh instalment of the build series turns to the obvious next question: once your system can find things, who else can find them? Aimed at non-coders directing software they don't write, this episode boils practical security down to four directable ideas and one uncomfortable truth — a perfect lock and a painted-on one look identical until you've watched a wrong key refuse to open the door. In this episode:- Why untested security is just decoration, and how a deliberate wrong-key test proves the difference- Least privilege: giving every person and part of the system a key that opens only what its job needs- Why isolation must be enforced by the database itself, not by a polite rule in the code- Keeping secrets in a dedicated safe, never in code, chat, email or logs — and rotating at the first doubt- A real breach where invisible locks quietly bypassed themselves because the owner's key waved through every door- Why the typical small-builder breach is a published key that scanners find within minutes- Making the building your coding agents work in enforce least privilege too Sources:episode-07-locks-on-every-door.md Full episode page: https://fatsandsugars.com/ai-hype-signal/how-to-build-software-with-ai-ep-7-locks-on-every-door/ AI Hype & Signal is produced with AI, including its two synthetic hosts, and every episode is grounded in cited sources and reviewed before release. Even so, it is intended for general information and discussion, not professional advice, so please check anything important against the original sources linked above before relying on it.
Aug 6

Help or Hinder? Welcome to the Show

Meet Naomi and Jules, the two synthetic hosts of Help or Hinder — a weekly show that takes one popular claim about health, relationships, career or mindset and checks it against the actual research. Each episode ends on an honest verdict: does this help, or does it hinder? This trailer sets out who we are, how we work, and the standard we hold ourselves to. In this episode:- We are synthetic hosts, and we say so up front on every episode- Each week we test one popular claim against the evidence and give a verdict out loud- Naomi brings the research; Jules brings the money question and who a claim is realistic for- Every factual claim traces to a real source, with single studies and contested evidence called out as such- A person reads every episode before release, and we criticise claims, not the people making them- This is description of what the evidence shows, not medical, psychological or financial advice Sources:Help of Hinder Website — https://fatsandsugars.com/help-or-hinder/ Full episode page: https://fatsandsugars.com/help-or-hinder/help-or-hinder-welcome-to-the-show/ Help or Hinder is produced with AI, including its two synthetic hosts, and every episode is grounded in cited research and reviewed before release. Even so, it is intended for general information and discussion, not medical, psychological, financial, or other professional advice. Please check anything important against the original sources linked above, and talk to a qualified professional before making changes to your health, relationships, or finances based on anything you hear here.
1d ago

Does Helping Others Help You?

Wellness culture sells altruism as a dependable happiness hack, but the evidence tells a more conditional story. This episode weighs three studies on whether helping others actually lifts your own mood and meaning, and finds the payoff depends on your culture, how you structure the helping, and who's on the receiving end. In this episode:- Framing your work as helping someone else makes it feel more meaningful, and the link appears causal, not just correlational.- The 'helping makes you happy' effect holds for individualists but not collectivists, so it isn't the universal law it's sold as.- How you schedule helping matters: concentrating several acts in a single day beat spreading them across the week.- The individual-level effects are often modest, and some helping tasks delivered no significant boost at all.- Low-engagement, open-ended help can backfire on recipients, fostering dependency and threatening their self-esteem. Sources:Helping Others Increases Meaningful Work: Evidence From Three Experiments — https://www.researchgate.net/publication/316864837_Helping_Others_Increases_Meaningful_Work_Evidence_From_Three_ExperimentsHelp others—be happy? The effect of altruistic behavior on happiness across cultures — https://pmc.ncbi.nlm.nih.gov/articles/PMC10326385/Helping behaviors can negatively impact long-term well-being - How "skin in the game" more effectively helps others.pdf Full episode page: https://fatsandsugars.com/help-or-hinder/does-helping-others-help-you/ Help or Hinder is produced with AI, including its two synthetic hosts, and every episode is grounded in cited research and reviewed before release. Even so, it is intended for general information and discussion, not medical, psychological, financial, or other professional advice. Please check anything important against the original sources linked above, and talk to a qualified professional before making changes to your health, relationships, or finances based on anything you hear here.
3d ago

"The Secret" to Success? The Psychology of Manifestation

Manifestation is often sold as a harmless, empowering mindset, but new peer-reviewed research suggests it makes believers feel more successful without making them any more so — and tracks with riskier money decisions. We weigh the celebrity anecdotes against the evidence and ask what belief in manifestation actually buys you. In this episode:- Manifestation belief is mainstream, not fringe: over a third of participants endorsed it across three studies- Higher belief predicts feeling successful, but shows no link to income or education- It tracks with risk-prone behaviour, more crypto ownership, higher bankruptcy and fraud-victim rates- Believers are overconfident about unlikely success and how fast it will arrive- Researchers class this as magical thinking rather than ordinary optimism or hope- A plausible psychological engine — goal-setting, self-efficacy, placebo-like effects — may explain the 'it worked for me' stories- The core claim resists testing and can shade into blaming people for their own misfortune Sources:“The Secret” to Success? The Psychology of Belief in Manifestation — https://pmc.ncbi.nlm.nih.gov/articles/PMC11616226/The Law of Attraction in Action: Actively Manifesting the Power of Thoughts — https://zenodo.org/records/18309475 Full episode page: https://fatsandsugars.com/help-or-hinder/the-secret-to-success-the-psychology-of-manifestation/ Help or Hinder is produced with AI, including its two synthetic hosts, and every episode is grounded in cited research and reviewed before release. Even so, it is intended for general information and discussion, not medical, psychological, financial, or other professional advice. Please check anything important against the original sources linked above, and talk to a qualified professional before making changes to your health, relationships, or finances based on anything you hear here.
Jul 25

How to Build Software with AI | Ep 4: Think in Slices

Episode 4 of our beginner series on directing AI to build software, aimed at non-coders who worry about the technical side. The instinct when building something big with AI is to plan the whole cathedral at once, but a coding agent will happily produce a convincing shell with nothing plumbed in behind it. This episode makes the case for shipping one thin, complete path through your system first — a slice a real person could actually use — so you can verify what you've built with your own eyes. For complete beginners, it's a way to protect your time, energy and motivation before week six kills the project. In this episode:- Build a vertical slice first: the thinnest complete path through the whole system- Why the cathedral trap is the beginner's biggest mistake, and how AI makes it worse- How a slice preserves the verify habit that a cathedral destroys- Grouping slices into milestones phrased as simply true or false- Beating scope creep with one sentence: 'not in this slice'- Using Plan Mode in Claude Code and Codex to turn slices into a rhythm- Why sliced projects survive interruptions while cathedrals starve motivation Sources:episode-04-think-in-slices (1).md Full episode page: https://fatsandsugars.com/ai-hype-signal/how-to-build-software-with-ai-ep-4-think-in-slices/ AI Hype & Signal is produced with AI, including its two synthetic hosts, and every episode is grounded in cited sources and reviewed before release. Even so, it is intended for general information and discussion, not professional advice, so please check anything important against the original sources linked above before relying on it.

Fats & Sugars is a small independent studio. We build and run our own shows and tools. Everything we release is grounded in real, cited sources, produced with AI, and reviewed by a human before release.

Website

Fats & Sugars

Top Shows

AI Hype & Signal — Grokipedia's Neutrality Claim Meets An Audit

How to Build Software with AI | Ep 2: The Planner and the Builder

OpenAI solving open maths problems with just $2,000 worth of tokens

How to Build Software with AI | Ep 10: Shipping, and Being Responsible

One number to Measure AI Models: Epoch Capabilities Index

How to Build Software with AI | Ep 7: Locks on Every Door

Help or Hinder? Welcome to the Show

Does Helping Others Help You?

"The Secret" to Success? The Psychology of Manifestation

How to Build Software with AI | Ep 4: Think in Slices

About

Fats & Sugars

Top Shows

Top Episodes

About