by Malcolm Murray

Enrico Fermi famously asked – allegedly out loud over lunch in the cafeteria – “Where is everybody?”, as he realized the disconnect between the large number of habitable planets in the universe and the number of alien civilizations we actually had observed.
Today, we could in a similar vein ask ourselves, “Where is all the AI-enabled cybercrime?” We have now had three years of AI models scoring better than the average humans on coding tasks, and four years of AI models that can draft more convincing emails than humans. We have had years of “number go up”-style charts, like figure 1 below, that show an incessant growth in AI capabilities that would seem relevant to cybercriminals. Last year, I ran a Delphi study with cyber experts in which they forecast large increases in cybercrime by now. So we could have expected to be seeing cybercrime run rampage by now, meaningfully damaging the economy and societal structures. Everybody should already be needing to use three-factor authentication.

But we are not. The average password is still 123456. The reality looks more like figure 2. Cyberattacks and losses are increasing, but there is no AI-enabled exponential hump.

So we should ask ourselves why this is. This is both interesting in its own right, as cyberattacks hold the potential of crippling our digital society, as well as for a source of clues to how advanced AI will impact the economy and society. The latter seems much needed at the moment, as there is significant fumbling in the dark. Just in the past month, two subsequent Dwarkesh podcasts featured two quite different future predictions. First, Daniel Kokotajlo and Scott Alexander outlined in AI 2027 a scenario in which AGI arrives in 2027 with accompanying robot factories and world takeover. Then, we had Ege Erdil and Tamay Besiroglu describing their vision, in which we will not have AGI for another 30 years at least. It is striking how, while using the same components and factors determining AI progress, just by putting different amounts of weight on different factors, different forecasters can reach very different conclusions. It is like as if two chefs making pesto, both with basil, olive oil, garlic, pine nuts, cheese, but varying the weighting of different ingredients, both end up with “pesto,” but one of them is a thick herb paste and the other a puddle of green oil.
Below, I examine the potential explanations one by one and how plausible it is that they hold some explanatory power. Finally, I will turn to if these explanations could also be relevant to the impact of advanced AI as a whole. Read more »


Sughra Raza. After The Rain. April, 2025.
Morality, according to this view, is more like taste, and in matters of taste I don’t expect others to be like me. This is of course incoherent since the very imperative to be non-judgmental is itself a moral demand, which must claim some level of objectivity since it is a rule that others are expected to follow. Judging others, according to non-judgmentalism, is something we ought not to do. It is presented as an objective moral rule.





On a hot summer evening in Baltimore last year, the daylight still washing over the city, I sat on my front porch, drinking a beer with a friend. Not many people passed by. Most who did were either walking a dog or making their way to the corner tavern. And then an increasingly rare sight in modern America unfolded. Two boys, perhaps ages 8 and 10, cruised past us on a bike they were sharing. The older boy stood and pedaled while the younger sat behind him.

If I were asked to name the creed in which I was raised, the ideology that presented itself to me in the garb of nature, I would proceed by elimination. It wasn’t Judaism, although my father’s parents were orthodox Jewish immigrants from the Czarist Pale, and we celebrated Passover with them as long as we lived in Montreal. It certainly wasn’t Christianity, despite my maternal grandparents’ birth in protestant regions of the German-speaking world; and it wasn’t the Communism Franz and Eva initially espoused in their new Canadian home, until the Molotov-Ribbentrop Pact put an end to their fellow traveling in 1939. Nor can I claim our tribal allegiance to have been to psychoanalysis, my mother’s professional and personal access to secular Jewish culture, although most of my relatives have had some contact, whether fleeting or intensive, paid or paying, with psychotherapy—since the legitimate objections raised by many of them to the limits of classical Freudian theory prevent it from serving wholesale as our ancestral faith, no matter the extent to which a belief in depth psychology and the foundational importance of psychosexual development informs our discussions of family dynamics.
About 45 years ago, psychiatrist Irvin Yalom estimated that a good 30-50% of all cases of depression might actually be a crisis of meaninglessness, an
Sughra Raza. Aerial composition, March, 2025.
