Tech as Teammate

In Surrender, Bono's beautiful and captivating account of the history of U2, he describes the profound change in the band's musical signature when they began working with producer Brian Eno in the early 1980s. The shift was most palpable in their recording of The Joshua Tree, a phase Bono describes as "a purple patch for the four of us, a time when I could hardly sleep for the excitement of the work I'd wake up in."

What stands out from Bono's description of their collaboration with Eno is the role of the recording studio. He says "Brian operated the studio as if it was a single instrument, and in his college of art we were still students."

Eno was prodigiously well educated in musical history, capable of describing how Irish melodies could be traced to North Africa and the Middle East. What galvanized the growth of U2, however, was Eno's mastery of studio technologies. He used the studio to capture and manipulate sound. He created atmospheric textures and new sonic territories. Through repeated experimentation, he pushed the band to forge a more expansive and distinctive voice. Human-machine interaction made them a much better version of themselves.

This might be a useful analogy for thinking about the role of AI in our professional lives today. Eno generated better 'code' than other producers because he was more knowledgeable about music, not just more conversant in technology. The power of the studio was proportional to his erudition.

As Ethan Mollick has noted in Co-Intelligence: Living and Working With AI, writers are often better at prompting AI because they can provide more interesting instructions, such as "end on an ominous note," "make the tone increasingly frantic," or "do this in the style of John McPhee."

Far from demonstrating the futility of human knowledge, these examples show how AI is holding humans to a higher standard of learning. For most of human history there was a master-servant relationship with technology. The wheel, the printing press, and the steam engine were all made to fulfill a narrower mandate. Large language models, on the other hand, can meet a much wider range of potential needs, and output quality is often correlated with input creativity.

In a recent Gartner keynote on the state of sales, speakers Alice Walmesley and Dave Egloff suggested technology has become less a tool than a teammate. That sounds about right to me. Tech is no longer subservient but collaborative. Humans have a heightened responsibility to bring value to our relationship with AI through judgment, learning, and taste.