Memorandum
- From
- Derek Weston via Fast Company
- Date
- Filed
- Business·5 min to read
- Re
Leo Aschenbrenner's AI hedge fund crashed after viral success
ReLeo Aschenbrenner's AI hedge fund crashed after viral success
Leo Aschenbrenner, once hailed as an AI prodigy, saw his Situational Awareness hedge fund collapse in July, exposing highly leveraged, correlated bets and raising questions about overconfidence in Silicon Valley.
Leo Aschenbrenner, a former OpenAI and FTX employee who became a celebrated voice on artificial intelligence, saw his hedge fund collapse in July after a relatively modest market correction. The fund, named after his viral essay Situational Awareness, had been up more than 1,000 percent in two years, but its bets turned out to be both highly correlated and highly leveraged. Forced to sell his public stocks at fire-sale prices to Ken Griffin, Aschenbrenner became the latest example of a familiar Silicon Valley pattern: confident prediction followed by a sharp reversal.
Aschenbrenner's rise was swift. Born in Berlin to a pair of doctors, he graduated from Columbia as valedictorian at 19. After brief stints at FTX and OpenAI, he published the essay that took Silicon Valley by storm. Theo Baker, writing in The Atlantic, described him as the « Nostradamus of AI », someone who offered clarity where others offered confusion. In the essay, Aschenbrenner wrote, « I can see how AGI will be built. » For a time, his results seemed to back up the bravado, even if he came across as cocky and somewhat eccentric.
The episode fits a broader argument about how new technologies are received. Philosopher Martin Heidegger, in his essay The Question Concerning Technology, described technology as a way of revealing, using the example of a hydroelectric dam that channels the energy of a river. When a new technology opens possibilities, the early period is marked by confusion about how to direct them, and those who speak most confidently gain an advantage. They attract investors, win coverage in the business press, receive invitations to major conferences and build large followings on social media.
Earlier technology booms show the same dynamic. Early web companies such as Webvan, eToys and reached huge valuations before crashing, and Enron, a highly successful energy trader, failed when it tried to repeat that success in markets such as water and bandwidth. The underlying ideas were not always bad; different versions of them later succeeded. The problem, the argument goes, is that every idea is wrong in some way, and a leader's job is to make it less wrong over time.
Research from the Federal Reserve Bank of St. Louis is cited as evidence that artificial intelligence is following a similar path. AI is helping productivity soar for some tasks, such as those related to computers and math, but for most work its impact has been far smaller. People closest to the tasks experiencing the boom can easily mistake their experience for a universal one, which gives them a certainty others do not share.
Psychologists call the tendency for one strong impression to shape judgments in other areas the halo effect. Phil Rosenzweig, in his book The Halo Effect, pointed to Cisco during the dotcom boom: when the company was flying high, its culture was praised as unmatched, with happy employees working long hours and loving every minute. When the market fell, the same culture was described as « cocksure » and « naive ». The pattern, critics say, looms especially large in tech, where success in one domain is often assumed to transfer to another.
Aschenbrenner's case illustrates the point. Observers impressed by his essay and his academic record assumed those skills would carry over to hedge fund investing, but they did not. Whatever the merit of his views on artificial intelligence, he failed to grasp basic principles of risk management. Social psychologist Jonathan Haidt argues in The Righteous Mind that many opinions reflect membership in a particular team. Aschenbrenner's techno-optimism, in this reading, signaled a specific identity that many found appealing, and the urge to display such identity can lead to group polarization and a purity spiral in which the most extreme views are rewarded.
3
