AI Temperature: The Hidden Setting That Controls Research Quality
Are you struggling to get reliable and repeatable results from AI in your professional research? This episode dives into the methodological rigour needed to turn large language models (LLMs) into trustworthy scholarly tools.Join our discussion with R&D Engineer James Sutherland to uncover the secrets of advanced prompt engineering, including why giving your chatbot a "role" and demanding reasoning drastically improves accuracy. We explore the critical concept of AI "temperature" (and how it affects deterministic results) and connect it to the crucial academic field of uncertainty quantification. Learn how to build a multi-dimensional approach to AI analysis, moving beyond the simple "answer" to a complete, explainable methodology essential for AI standards and regulation.
--------
26:47
--------
26:47
From King Charles to the Louvre: Who Controls AI & Why Passwords Fail
Are you trying to understand the geopolitical and technical reality of the AI revolution? You're in the right place. This episode unpacks King Charles's symbolic warning to Nvidia's CEO and explains why Jensen Huang's company is the unrivalled $4 Trillion power behind modern AI.Join Lucian, Emily, and Mark as they dive deep into the hardware dominance of Nvidia's GPUs and the CUDA platform, exploring the ethical and societal risks of concentrated AI control. We discuss the massive computational cost of LLMs and the push for manufacturer guardrails. Plus, we pivot to a stunning cyber security failure: how the Louvre Museum's $1.2 million heist revealed the trivial password "Louvre," and what this massive security gap teaches us about cyber security across all industries.
--------
41:22
--------
41:22
AI Robot Caregivers, DNS Failures, and the Uncanny Valley
Welcome to the essential discussion on the biggest tech debates of the week! If you’re trying to understand the terrifying future of humanoid AI robots and the real cause of global cloud outages, you're in the right place.This week, we dive deep into the controversial launch of the 1X Neo AI robot, debating its potential to revolutionize (or ruin) disability and elderly care, the ethical pitfalls of emotional exploitation, and the unsettling reality of the uncanny valley. Then, we break down the critical Microsoft Azure and AWS cloud outages, explaining why the obscure Domain Name System (DNS) is the internet's central "kill switch" and how this common failure mechanism poses a massive cybersecurity risk to our connected world. Plus, a quick detour into AI-generated art in music!
--------
36:09
--------
36:09
How Blockchain Stops Deepfakes: Manny Ahmed, CEO of OpenOrigins Explains
Welcome to the reality check on the internet's biggest threat: deepfakes and the total loss of trust in digital media. If you're wondering if we can still tell what's real and what's fake online, you're in the right place.In this vital discussion, Dr. Manny Ahmed, founder of Open Origins and visiting researcher at Cambridge, explains why the deepfake 'arms race' of detection is over, and introduces the only sustainable solution: provenance. Learn how Open Origins uses secure sourcing (3D depth mapping on cameras) and archive anchoring (blockchain for historical footage) to create an unforgeable audit trail that definitively proves media authenticity. This paradigm shift is not just protecting news organizations like ITN, it’s fighting for our shared sense of reality and the future of democracy in a bot-dominated digital world.
--------
34:45
--------
34:45
Mandatory ID and Watermarks: Big Tech's Ethical Failures
Is the government forcing a Digital ID on a public that doesn't want it, while Big Tech simultaneously abandons all safeguards against Deepfakes? The lines between technology, ethics, and control are vanishing.We dive into two monumental issues. First, the Sora AI Controversy: The visible watermark was useless and immediately bypassed. In a shocking move, OpenAI's Sam Altman is pushing for "societal co-evolution," forcing the public to "adapt to a world of synthetic video." We argue this is "naive consequentialism," shifting the blame for scams and deepfake fraud onto the user. Second, the UK Digital ID Plan: Mandatory for "Right to Work" checks, this scheme mirrors past ID card failures and India's Aadhaar system. We discuss the steep drop in public support, the fear of mass surveillance and mission creep (exclusion of the vulnerable, transgender rights), and why politics, not practicality, is driving the current debate.
"**The Internet, Seriously**" dives headfirst into the wild and wacky world of tech, news, and politics, all from the unique perspective of the internet itself! Join our trio of "Internet Governance and techie specialists" – Emily Taylor, Lucien Taylor, and Alice Taylor – as they bring their combined 50+ years of internet industry smarts to the mic.
---
Ever wondered about the secret life of robot lawnmowers, or pondered the existential musings of the Wiwiwi Cat? We've got you covered. From the nitty-gritty of cybersecurity and the latest EU regulations to the mind-bending implications of AI.
UK