AI tools are not human colleagues or friends warns expert

Gary Bolles, chair for the Future of Work at Singularity University, argues that personifying AI software risks eroding human relationships and workplace dynamics.

Key Points:

• "AI" is a misleading marketing term - The label oversimplifies complex technologies like generative AI and LLMs, doing tech companies' marketing for them
• AI applications are tools, not colleagues - They should be viewed as incomplete products with arbitrary limitations, not human equivalents
• Social AI creates false relationships - Apps like Replika ("the AI companion who cares") exploit human needs for affirmation while removing healthy relationship friction
• Serious limitations remain - Even major companies like OpenAI and Apple produce flawed outputs (e.g., ChatGPT planning a trip to an ocean baseball field)
• Unchecked use risks disasters - Autonomous agents could misuse data, make unauthorized purchases, or even instruct illegal activities
• Personification erodes humanity - The "AI employee" narrative devalues human workers and blurs important boundaries

"Our brains are far more hackable than we want to believe," Bolles notes, emphasizing this isn't anti-technology sentiment but a call for clearer boundaries.

While acknowledging AI's valuable applications, Bolles warns against treating software as coworkers, employees or friends. His book The Next Rules of Work explores these themes further.

Privacy Policy