Jagged Fascination

💡 The concept of "jagged frontier", introduced about two years ago (Fabrizio Dell'Acqua, Edward McFowland III, Ethan Mollick, Dr. H /Hila Lifshitz (Hán, X也), Katherine Kellogg, Saran Rajendran, Lisa Krayer, PhD, François Candelon and Karim Lakhani) and popularized by the extraordinarily energetic Ethan Mollick, really hit the spot at the right time to provide a simple metaphor for AI's uneven capabilities, where advanced models excel at some tasks (like creative writing) but fail at seemingly simpler ones (like basic math). This unevenness creates an unpredictable landscape of superhuman peaks ⛰️ and valleys in performance, requiring humans to experiment to find AI's limits and strengths for effective use, as it advances constantly with new models. And there is no more relentless experimenter than Ethan Mollick.

One reason for its success might be that it can be visualized (figure 1 in the original report, which I added here for reference) and its evolution extrapolated.

I had not realized until recently the extent to which it could stimulate people's imaginations 🧠 ! And thereby serve as a visual representation of possible futures. A message on X by Tomás Pueyo introduced a couple of variants and triggered a wave of additional takes from others. They are all different and each provides a unique perspective on where AI is going. Here is a small gallery, head for the X thread for commentary and proper attribution (link in comments).

Enjoy!

Additional: