By Lee Flanagan
Weekly Roundup · 2026-07-13 to 2026-07-17
Twenty-six former Meta employees are suing the company, claiming its AI-driven layoff tools penalized them for taking protected leave. Meanwhile, Bending Spoons hired 286 people from 800,000 applicants and called it a hiring strategy. This week’s stories share a pattern: systems that look like rigor but may be performing something else entirely.
'The AI did it' is not a legal defense
The Meta lawsuit alleges that algorithmic performance tools penalized employees for taking medical and parental leave, then flagged them for layoffs. For TA leaders using algorithms in hiring or workforce decisions, the lesson is blunt: you own the outcome, not the vendor. If your tool produces disparate impact, saying you did not understand how it worked will not protect you in court or in front of your board.
India's AI talent gap is also an assessment gap
India faces a shortage of 600,000 AI professionals by 2027, driven by curriculum lag and faculty gaps. But the piece surfaces a second problem: while 90% of engineering graduates claim AI skills, only 23% are considered AI-native by employers. That gap suggests the supply problem is compounded by an assessment problem. If you cannot tell who actually has the capability, you cannot hire your way out of scarcity.
A 0.04% acceptance rate is not proof your hiring works
Bending Spoons processed 800,000 applications to hire 286 people, then published the ratio as evidence of rigor. Extreme selectivity is theatre, not strategy. It proves you attracted volume, not that your selection process identifies the right people. Rejection rate is not the same as predictive validity. A funnel that rejects 99.96% of applicants may be rigorous, or it may simply be expensive and slow.
The through line this week is accountability. Algorithms let us scale decisions, but they also let us distance ourselves from those decisions. The question is whether that distance is operational efficiency or moral outsourcing. If you cannot explain why your system made a given choice, you do not have a system. You have a liability with a dashboard.