Uploaded October 2025 | Updated September 2026, 2 hours ago
What if the AI industry's obsession with building ever larger language models is actually making agents worse at their jobs? This presentation takes you on a journey from the early days of language models to today's trillion parameter behemoths, revealing why these impressive giants consistently stumble when tasked with real-world agent scenarios. We'll explore a counterintuitive solution small, strategically fine tuned models under 20B parameters—and showcase benchmarks that demonstrate how focusing on specialisation over scale might just revolutionise how we build practical AI agents.
Exchange events are brought to you by LinuxRecruit, building the next wave of Tech Unicorns in Europe.
Whether you're hiring or exploring your next opportunity, get in touch. linuxrecruit.co.uk/jobs
What if the AI industry's obsession with building ever larger language models is actually making agents worse at their jobs? This presentation takes you on a journey from the early days of language models to today's trillion parameter behemoths, revealing why these impressive giants consistently stumble when tasked with real-world agent scenarios. We'll explore a counterintuitive solution small, strategically fine tuned models under 20B parameters—and showcase benchmarks that demonstrate how focusing on specialisation over scale might just revolutionise how we build practical AI agents.
Exchange events are brought to you by LinuxRecruit, building the next wave of Tech Unicorns in Europe.
Whether you're hiring or exploring your next opportunity, get in touch. linuxrecruit.co.uk/jobs
