When AI sounds right but is wrong – what “strawberry” reveals about the limits of today’s AI

Academic analysis from LSE warns about the dangerous gap between AI fluency and reliability, using the 'strawberry' example (likely referring to how LLMs can confidently give wrong answers about counting letters). The piece argues that AI systems are being deployed faster than expected in critical processes despite fundamental limitations, creating structural risks when fluent-sounding but incorrect outputs are trusted.