"Finally, we are very concerned that this GPT could be unaligned with humans. This would be bad. We want this to be a nice GPT that deeply loves all humans and is always considerate and helpful. Thanks"

→ View original post on X — @karpathy