Here's what GPT-5 critics don't understand: Current models already outperform humans in fair competitions. Their impact is limited by context, cost, and affordances.
316