AI will support people in the work they do every day
OpenAI ha dado un paso inusual esta semana al publicar un estudio que nombra, con precisión y datos, las 44 ocupaciones donde la inteligencia artificial ya puede realizar tareas de valor económico real. Lo que durante años fue debate abstracto sobre el futuro del trabajo se convierte ahora en una lista concreta: abogados, enfermeras, desarrolladores de software, asesores financieros. La empresa enmarca el hallazgo como una herramienta de diálogo, pero el gesto de nombrar es, en sí mismo, un umbral cruzado.
- OpenAI publica el estudio GDPval y hace lo que siempre había evitado: señalar con nombre propio los empleos que su propia tecnología ya puede desempeñar.
- La lista abarca el sistema nervioso de la economía moderna —atención al cliente, derecho, salud, ingeniería, finanzas— y no como vulnerabilidades futuras, sino como capacidades presentes y medibles.
- Voces como las de Mo Gawdat y Sam Altman ya anticipaban esta disrupción; el informe convierte sus advertencias en evidencia respaldada por los propios creadores de la tecnología.
- OpenAI elige con cuidado sus palabras: la IA 'apoyará' a los trabajadores, no los reemplazará, una distinción que reconoce la tormenta sin pronunciar su nombre.
- La incertidumbre no ha desaparecido, pero se ha vuelto más estrecha: ya no se debate si la IA puede hacer trabajo de conocimiento, sino cuán rápido llegará la adopción y qué ocurrirá con quienes la vivan.
OpenAI publicó esta semana el estudio GDPval, un informe que identifica 44 ocupaciones donde ChatGPT y sistemas similares ya pueden realizar tareas de valor económico real. La lista no es especulativa: los propios investigadores de la compañía probaron sus modelos contra trabajo del mundo real y encontraron capacidad funcional y medible. Un agente inmobiliario puede usar la herramienta para diseñar materiales de venta. Una enfermera, para evaluar imágenes dermatológicas. La tecnología no espera en el horizonte; ya opera en el presente.
El momento tiene peso propio. Figuras como Mo Gawdat, exdirector comercial de Google X, y Sam Altman, CEO de OpenAI, han advertido públicamente que la atención al cliente será el primer sector en transformarse radicalmente. El informe GDPval convierte esas advertencias en datos: abogados, farmacéuticos, desarrolladores de software, asesores financieros y trabajadores sociales figuran entre los perfiles analizados.
Sin embargo, OpenAI cuida su lenguaje con deliberada precisión. El informe no habla de sustitución sino de 'apoyo' a los trabajadores, y se presenta como instrumento para sostener conversaciones basadas en evidencia, no como pronóstico de desempleo masivo. Es una distinción que reconoce la magnitud de lo que se avecina sin nombrarlo directamente.
Lo que hace singular este momento es el tránsito de lo abstracto a lo concreto. Durante años, el debate sobre la IA y el trabajo fue teórico. Ahora hay una lista. Cuarenta y cuatro ocupaciones. Tareas reales. Rendimiento medible. La pregunta ya no es si la inteligencia artificial puede hacer trabajo de conocimiento. La pregunta es a qué velocidad ocurrirá, qué sectores se moverán primero, y qué pasará con las personas cuyas habilidades, de pronto, se vuelven abundantes.
OpenAI released a report this week that does something the company has been careful to avoid until now: it names the jobs that artificial intelligence can already do. The study, called GDPval, identifies 44 occupations where ChatGPT and similar AI systems could meaningfully change how work gets done—or whether humans do it at all.
The list reads like a map of the modern economy's nervous system. Customer service representatives. Software developers. Lawyers. Nurses. Pharmacists. Financial advisors. Social workers. Private detectives. Industrial engineers. These are not hypothetical vulnerabilities or distant possibilities. OpenAI's own researchers tested their AI against real-world tasks in these fields and found that the systems could already handle specific, economically valuable work. A real estate agent could use ChatGPT to design a sales brochure. A nurse could use it to evaluate images of skin lesions. The capability is there, functioning, measurable.
The timing matters. Industry voices have been growing louder about what's coming. Mo Gawdat, who ran Google X's commercial operations, has said it's likely that AI will eventually replace your job. Sam Altman, OpenAI's CEO, recently predicted that customer service would be the first role to fall. These aren't fringe warnings anymore—they're coming from people running the companies building the technology. The GDPval report, in that context, feels like OpenAI putting numbers behind what everyone already suspects.
But the company is being careful with its language. OpenAI doesn't say these jobs will disappear. Instead, the framing is that AI will "support people in the work they do every day." It's a deliberate choice of words, one that acknowledges the disruption without naming it directly. The report itself is positioned as a tool for having "evidence-based conversations about the future of AI" rather than as a prediction of mass unemployment. OpenAI's researchers explain that studies like GDPval help them track how their models improve over time and ground discussions in data rather than speculation.
What makes this moment significant is the shift from abstraction to specificity. For years, the AI conversation has been theoretical—what could happen, what might change, how society should prepare. Now there's a concrete list. Forty-four occupations. Real tasks. Measurable performance. The uncertainty hasn't disappeared, but it's narrowed. The question is no longer whether AI can do knowledge work. It's how quickly adoption will happen, which sectors will move fastest, and what happens to the people whose skills are suddenly abundant.
Citações Notáveis
Evaluations like GDPval help ground conversations about AI's future in evidence rather than speculation, and allow us to track how our models improve over time— OpenAI researcher, cited in company blog post
AI will support people in the work they do every day— OpenAI, in GDPval report framing
A Conversa do Hearth Outra perspectiva sobre a história
Why did OpenAI release this list now? They've been cautious about the job displacement question for years.
They're trying to control the narrative by being the ones who measure it first. If you publish the data yourself, you shape how people interpret it. You get to say "support" instead of "replace."
But the list is pretty damning. Forty-four jobs. That's not a small number.
It's not. But notice what they're measuring—not whether the jobs disappear, but whether AI can do specific tasks within those jobs. A nurse's job is more than reading skin images. A lawyer's job is more than document review. The framing leaves room for both the technology and the employment to coexist.
For how long, though? If the technology keeps improving, doesn't the "support" eventually become "replacement"?
That's the question nobody wants to answer directly. OpenAI certainly doesn't. But the fact that they're publishing this at all suggests they know the answer is probably yes, and they'd rather be transparent about the capability than be caught hiding it later.
So this is damage control?
It's more like controlled disclosure. They're saying: here's what we can do, here's the evidence, now let's talk about what it means. It's better than silence, but it's also not the same as saying what they actually think will happen.