A plan that was approved and then abandoned
On December 9, 2019 — the last full day of Mauricio Macri's presidency — Argentina's Secretariat of Science, Technology and Productive Innovation signed off on the Plan Nacional de Inteligencia Artificial, known as ArgenIA. The document laid out six strategic axes (talent, data, infrastructure, R&D, public-sector adoption, and ethics/regulation) and proposed a governance architecture: an executive table, an innovation lab, a scientific committee. It was ambitious on paper. It also arrived with no budget line, no administrative structure, and a government about to leave office the following day. Six and a half years later, a new academic study says that plan never left the drawing board — and that the fragmented, uncoordinated response to AI since then is now Argentina's single biggest structural risk.
What the new study actually found
Researchers from Universidad Austral, CONICET, and Universidad Católica de Santa Fe published the Estudio Prospectivo IA 2035 in early June 2026, using the Delphi method — structured, anonymous, multi-round consultation designed to converge expert opinion — to survey 67 specialists spanning government, academia, and industry (Universidad Austral). The headline finding: 83% of those experts identified fragmentation and lack of articulation between the three sectors as the primary obstacle to scaling AI solutions nationally (Espacio Tech). The study places Argentina at 0.47 on the IMF's AI Preparedness Index — behind Chile (0.59), Uruguay (0.55), and Mexico (0.53), among the weaker performers in Latin America's upper tier (Buenos Aires Times). The researchers' central recommendation is not a new regulatory regime — it is reviving the dormant 2019 initiatives through genuinely multisectoral governance, alongside goals like expanded connectivity, federal labor-reconversion policy, teacher training, and SME tech-transfer incentives.
The case for coordination — steelmanned
There is a real argument here, and it deserves to be taken seriously rather than waved off as bureaucratic reflex. AI adoption that proceeds sector-by-sector, agency-by-agency, without any shared data standards, workforce pipeline, or shared risk framework tends to produce duplicated effort, incompatible systems, and — critically for a federal country with deep regional inequality — benefits concentrated in Buenos Aires while the interior falls further behind. Argentina isn't starting from zero: a Mesa Interministerial de Inteligencia Artificial convened at Casa Rosada in November 2023 under the outgoing government, bringing together Strategic Affairs, Labor, Health, Science, Defense, Security, Economy, and Foreign Affairs to look at exactly these cross-cutting risks (Argentina.gob.ar). That body, too, appears to have produced no durable institutional output. Two governments, two attempts at coordination, two dead ends — that pattern is exactly what should worry anyone who wants Argentina to actually capture AI's productivity gains rather than watch them accrue elsewhere.
Why the fix isn't a new regulator
Where the study's diagnosis is right, the temptation it invites — stand up a large new AI regulatory bureaucracy — would be the wrong prescription. The Milei government has taken the opposite tack: a bill now before the Senate, championed by Deregulation Minister Federico Sturzenegger, would create "automated societies" — corporate structures that can operate through AI agents without human directors — explicitly rejecting what Milei calls "the deadly hand of premature regulation" (Infobae). Technology-law specialist Alan Daitch has already flagged that the bill doesn't resolve who bears legal responsibility when an AI-run entity causes harm — a real gap, since Argentine corporate law still requires an identifiable person or entity to answer for damages. That critique is worth heeding on its own terms. But it is a separate problem from the coordination gap the Delphi study surfaces, and conflating the two would be a mistake.
The two findings actually point toward a coherent, proportionate answer: Argentina doesn't need a heavy new AI regulator to fix fragmentation, and it doesn't need to abandon liability principles to attract investment. What ArgenIA's 2019 blueprint got right was proposing a lightweight coordinating body — not a rulemaking agency — to align data standards, workforce training, and public-sector procurement across ministries and universities, while leaving substantive regulation minimal and targeted (Daitch's liability point is the kind of narrow fix that belongs in corporate law, not a horizontal AI statute). A coordination table with actual budget and a mandate that survives a change of government would address the 83% concern directly, without slowing the deregulatory experiment the current administration is running with its Senate bill. The failure mode to avoid is doing neither: leaving coordination dormant for a third administration in a row while a liability-light corporate bill moves ahead unchecked. Argentina has written this plan down twice already. The missing ingredient was never analysis — it was follow-through.