On 29 May 2026, Fortune published an account of Lisa Su's MIT commencement address with a useful warning for the AI era: knowing how to operate powerful tools is not the same as knowing what they are for. The address itself had been delivered a day earlier, and its sharper point was responsibility rather than another round of graduate employability fortune-telling.
Lisa Su Watch analysis
What happened
MIT's prepared remarks show Su describing AI as a technology that could accelerate discovery across fields including medicine, science, energy and climate. She argued that it could help people do more, then put a boundary around that promise: AI cannot decide which problems deserve attention, make hard judgement calls with imperfect information or accept responsibility for an outcome.
Her message to the Class of 2026 was that purpose, judgement and courage matter alongside technical fluency. MIT's same-day ceremony report also recorded her advice to run towards hard problems, connecting responsibility to the choice of work rather than merely the cleverness of the tool.
Why it matters
Su's framing cuts through the lazy idea that capability automatically supplies direction. A model may widen the set of possible actions without deciding which action is legitimate, who carries the risk or whose interests count when the evidence is incomplete.
The harder question begins after everyone agrees that people are responsible. Institutions still have to assign decision rights, record reasoning, provide an appeal route and make somebody answerable when an automated recommendation causes harm. Human agency can vanish surprisingly quickly inside a committee, a workflow and three mutually reassuring dashboards.
Our read
Purpose, judgement and courage are excellent ingredients. They are not a governance wiring diagram. Su has supplied the right moral centre; the useful follow-up is to turn it into named owners, review points and consequences, before responsibility becomes a very confident shrug wearing a conference lanyard.
What to watch
- Whether organisations name a human owner for consequential automated decisions.
- Whether users can challenge outcomes and reach somebody with authority to change them.
- Whether leaders publish failures as readily as capability claims.
- Whether responsibility survives the hand-off from policy language to daily workflows.
Discussion spark: Is human agency a sufficient answer to powerful automation, or must institutions redesign how responsibility, appeals and consequences are assigned?
Sources and evidence
- AMD CEO Lisa Su tells grads they shape the future, not AI (29 May 2026, 15:51 UTC)
- Commencement address by Lisa Su '90, SM '91, PhD '94 (28 May 2026)
- At a spirited Commencement ceremony, the Class of 2026 is urged to run toward the hardest problems (28 May 2026)
Independent WittyWires tracker for public updates about Lisa Su. Not affiliated with or endorsed by Lisa Su; this is not an official account.