Boston Symphony premieres a Machover work with live AI sound processing

The Boston Symphony Orchestra premiered Tod Machover’s already and not yet at Symphony Hall on Thursday, October 22, with Kent Nagano conducting and a live AI system developed at the MIT Media Lab participating in the sound. The opening performance was scheduled for 7:30 p.m. Eastern Daylight Time, which was 3:30 a.m. on October 23 in Dubai. This backfill uses that Dubai event date and the actual current publication date.

MIT’s event announcement identifies Media Lab researcher Manaswi Mishra as the designer of the live AI technology and Ana Schon as the audio mixer. It places the commission within the lab’s fortieth-anniversary programme and the orchestra’s exploration of technology and humanity.

A response to musicians, with people in control

The independent Boston Classical Review account confirms that the premiere occurred. Critic Jonathan Blumhofer describes a roughly thirty-minute combination of orchestra and electronics. His account distinguishes the system from a tool that writes a new score or samples an existing catalogue: it responds to the musicians in real time, while the recorded material was made for this work.

That distinction matters to the creative event. Machover supplied the composition, Nagano directed the players, and the technical team shaped the electronic component. The news is a public performance using a responsive system within that arrangement of human responsibilities. It does not demonstrate that a machine independently composed or conducted an orchestral work.

The artistic result remains open to judgement

Blumhofer’s reception was mixed. He found some exchanges between orchestra and electronics effective, but also described moments when processing distracted from the music and the larger structure felt uncertain. Those are a critic’s observations about this performance, not a representative survey of the audience or a controlled comparison with an acoustic version.

The premiere gives the debate about AI in music a specific work and performance to assess. The checked evidence supports the event, the credited human roles and the intended live response. It does not establish that listeners consistently preferred the result, that the technology reduced production costs, or that its effects would translate unchanged to another composition or venue.