What’s your take on the latest trend in Codex toward overoptimizing its reasoning? It sometimes produces weird, cut-off sentence beginnings like “I…,” “I am…,” or “I seem…” without developing the thought or explaining itself clearly. Is this a sign that future models will increasingly prioritize token efficiency at the expense of clarity and natural communication?
Loading post…
Join Marsel to reply and react. Browsing is free — sign in or request access to take action.