Most AI commentary is written, whether consciously or not, for a reader who already broadly agrees with the piece's general orientation-optimistic commentary reassures already-optimistic readers, skeptical commentary validates already-skeptical ones. A useful discipline that's rare in practice is writing explicitly for the reader who currently holds the opposite view, and being honest about whether the argument as constructed would actually be likely to move them.
Why writing for the already-convinced is the easy path
An argument built for a sympathetic reader can rely on shared assumptions and doesn't need to address the strongest version of the opposing view, because the audience isn't holding that view in the first place. This produces commentary that feels satisfying to write and to read within its intended audience, but that does little to actually shift anyone's thinking, since the readers who most need to encounter a strong counterargument aren't the ones the piece was built to reach.
What writing for the skeptical reader requires
Commentary genuinely aimed at changing a skeptical reader's mind has to engage seriously with the strongest version of their existing position, rather than a weaker version that's easier to dismiss, and has to acknowledge genuine tradeoffs and uncertainty rather than presenting the argument as one-sided. This is harder to write and often less immediately satisfying, both for the writer and for an already-sympathetic reader who might prefer a more triumphant framing.
A useful test before publishing opinion
A reasonable check for any AI opinion piece is to ask: would someone who currently holds the opposite view, reading this in good faith, find anything here that actually engages with their strongest reasoning, or does the piece only address a weaker version of their position? If the honest answer is the latter, the piece is more likely to entrench existing views than to inform anyone genuinely undecided.
Why this standard matters for a genuinely uncertain beat
AI's business, technical, and policy implications remain genuinely contested and uncertain enough that commentary should be trying to inform real disagreement, not simply provide validation to whichever side a given reader already occupies. Naming the skeptical reader explicitly is a useful discipline for actually attempting that.