Discounting the short-term risks of AI, from phishing to propaganda, simply because artificial superintelligence is not here yet will leave society ill-prepared
Superintelligence may or may not be imminent. But there's a lot to be worried about, either way. — Is AI going to kill us all?
Context & Ripple Effects
This argument lands mid-debate about whether AI danger is a far-future problem. Two weeks earlier, The New York Times framed the choice starkly: humanity must either accelerate its adaptation or reach a collective, enforceable decision to slow development, and Vox made the parallel case that slowing down is defensible even if the arms-race logic makes AI feel inevitable.
What this piece adds is a reframing: the fixation on artificial superintelligence — and on whether it is imminent — crowds out attention to harms already deployable at scale, from phishing to fraud to propaganda. The later record bears this out: by early 2025 an ex-OpenAI safety researcher warned that labs racing to AGI can cut corners on alignment, pushing everyone else to speed up too.
First-order effects
- The immediate audience is policymakers and security planners: treating phishing, fraud, and propaganda as pre-AGI footnotes deprioritizes defenses against attacks that require no breakthrough to scale.
Second-order effects
- As labs compete on speed rather than caution — the dynamic the ex-OpenAI researcher describes — near-term misuse risk compounds faster than oversight, which is exactly what security researchers flag when they note global regulatory frameworks are lagging behind assessments of AI's destructive potential.
Third-order effects
- If the pattern holds, the industry's structure splits into two uncoordinated tracks — capability racing inside labs and reactive harm mitigation outside them — forcing the choice the Times laid out: either adaptation infrastructure catches up on its own or a collective, enforceable slowdown becomes the only remaining lever.
The trend: The AI risk conversation is splitting into an existential-risk camp and a near-term-harm camp, and the gap between deployment pace and governance is widening as the latter gets discounted.