Jürgen Schmidhuber

Jürgen Schmidhuber (born 1963) is the German computer scientist the New York Times once suggested AI might someday call “Dad.” As scientific director of the Swiss lab IDSIA, he co-developed the long short-term memory network (LSTM) with his student Sepp Hochreiter in the 1990s — the architecture that later powered speech recognition and translation on billions of phones. Since 2021 he has led the AI initiative at Saudi Arabia’s KAUST, and he has spent years publicly insisting that the deep-learning trio of Hinton, Bengio and LeCun built their fame on foundations his lab laid first.

The refusenik of the doom letters

When the 2023 wave of AI alarm crested — the Future of Life Institute’s pause letter, the Center for AI Safety’s extinction statement — Schmidhuber conspicuously declined to sign. He told the Guardian in May 2023 that an AI arms race between governments, universities and companies made a pause impossible, that the technology’s rise “should not be feared,” and that counter-AI would police misuse. He has argued many doom-warners are “just seeking publicity,” and told Forbes the same year that his life’s work will not lead to dystopia.

His optimism runs cosmic: Schmidhuber has long predicted self-improving machines that surpass human intelligence and eventually expand into space, framing humanity as an important step — but not the last one — in the universe’s evolution toward complexity. In his telling, superintelligent AIs will be largely uninterested in humans rather than hostile to them.

Critics find his cheer too breezy and his priority battles exhausting, but few dispute the record: the tools he built in obscurity in the 1990s became the plumbing of the modern AI economy long before it was fashionable.