AI Models Counting Letters: A Playful Exploration

In episode 14 of The AI Fix, Graham makes an apology, Mark wonders if suicide drones have second thoughts, people pretend to be robots, and some researchers prove that all you need for an AI to generate a somewhat usable version of the computer game Doom out of thin air is to already have a fully-working copy of the computer game Doom.

Graham learns how to escape from a police sniffer elephant, an AI-generates a smell with no odour, and Mark explains why the world's best LLMs think there are two Rs in "strawberry".

Episode links:

• Women will be having more sex with robots than men by 2025.
• The Vatican wants a moratorium on killer robots.
• Kim Jong Un inspects new ‘suicide drones,’ urges incorporation of AI.
• OpenAI in talks for funding round valuing it above $100 Billion.
• Women are less interested in AI than men.
• New AI model can hallucinate a game of 1993’s Doom in real time.
• GameNGen playing Doom.
• Chinese robot conference spotlights changing face of humanoids.
• Humanoids (robots or otherwise) perform at World Robot Conference.
• Led by a former Google researcher, this company wants to give computers a sense of smell.
• Computers are learning to smell.
• OpenAI Tokenizer.
• Introducing Meta Llama 3: The most capable openly available LLM to date.
• The Problems with Tokenization in LLMs.

The AI Fix

The AI Fix podcast is presented by Graham Cluley and Mark Stockley.

Learn more about the podcast at theaifix.show, and follow us on Twitter at @TheAIFix.

Never miss another episode by following us in your favourite podcast app. It's free!

Like to give us some feedback or sponsor the podcast? Get in touch.