Keep an eye out, that Chameleon is shifty.

Comparison:

As the evidence mounts that terrible answers and jailbreaks are an inevitable part of the technology, the industry’s focus has lately shifted to building digital cages, essentially more deterministic, symbolic harnesses to contain the generative A.I. engine and check its results.

Zeynep Tufekci writing in the New York Times suggests that our worry over AI stealing jobs, while not unfounded, is overblown in much media coverage. How then to control jailbreaks and other AI mal-behavior? Digital cages. Hmmmm . . . digital cages, could such a cage consistently control, or as Tufekci puts it, “harness“ AI? Stay tuned. Unsurprisingly other forms of control--guardrails, boundaries, barriers show up in this opinion essay. Are our metaphors for control and containment adequate for our hyper-speed, quantum, digital times? What other comparisons would you use concerning keeping AI in check?

Context:

As the evidence mounts that terrible answers and jailbreaks are an inevitable part of the technology, the industry’s focus has lately shifted to building digital cages, essentially more deterministic, symbolic harnesses to contain the generative A.I. engine and check its results. Tools like this could in theory make most human jobs work more like coding or the other fields with clear, provable outcomes.

Citation:

Tufekci, Zeynep. “The One Very Simple Reason A.I. Won’t Steal All Our Jobs.“ New York Times, 30 June 2026. Web.

(Graphic image courtesy of GoogleLabs, July 2026.)