Saying AI does funky things from time-to-time is hardly an extraordinary claim. Meanwhile, an individual injecting bias without the hack's true purpose being o

Saying AI does funky things from time-to-time is hardly an extraordinary claim. Meanwhile, an individual injecting bias without the hack's true purpose being obvious to co-workers would be difficult. The more subtle and circumstantial the hack, the more likely it will be broken/ignored by someone else's changes. A hack is still possible, but I'd want evidence of intentional bias before entertaining the claim.