Sam Altman, CEO of OpenAI, said at a Sydney conference on May 26, 2026 that “AI will not lead to a so-called ‘jobs apocalypse’“, effectively revising his earlier prediction. He directly admitted, “I was wrong on this point“. It’s the first public self-correction by the creator of ChatGPT regarding AI’s job-market impact predictions.

This article synthesizes Hankyung’s 2026.05.26 Sydney coverage and analyses from McKinsey and MIT Tech Review to cover Sam Altman‘s Before/After prediction shift, 5 reasons AI unemployment lagged expectations, global enterprise adoption (HSBC·Amazon·StanChart·CBA), 5 human-relationship domains hard to replace, 5 actually-threatened roles, 5-year 3-scenario outlook, and 5 investment strategies. Original Hankyung article at Hankyung.

Sam Altman from 2022~2024 (right after ChatGPT launch) frequently spoke in a “AI will rapidly replace white-collar jobs” tone. He argued 30~50% of jobs would shift within 5 years and that social safety nets like UBI (universal basic income) might be necessary.

But in Sydney in May 2026, he publicly corrected: “I was wrong on this point”. A self-correction by the CEO of the AI industry’s #1 company on his core claim is rare and sends a strong signal to markets and policymakers that AI impact is gradual. This matters for both AI infrastructure firms and human-relationship industries.

The slower-than-expected pace of AI unemployment isn’t due to a single factor. Need for human relationships, regulation/liability, retraining/transition pace, union negotiations, cost/integration difficulty compound.

The biggest is need for human relationships. Customer/colleague trust, subtle emotional handling, and ethical judgment cannot yet be fully replaced by AI. Second is regulation/liability in legal/medical/financial sectors. Lack of clear legal frameworks for who bears responsibility when AI makes wrong decisions slows adoption.

Global cases cited in Hankyung’s report: HSBC, Amazon, Standard Chartered (StanChart), and CBA (Commonwealth Bank). Adding Goldman Sachs and JPMorgan to make 6 case studies, headcount reductions are 5~18% with simultaneous AI-related hiring, so “net reductions” are smaller than predicted.

Amazon is most aggressive (~-18%). Call-center and tier-1 customer support automation reduced headcount, but AI/cloud hiring is also active (+8%), so net change is about -10%. HSBC -12%/+4% nets -8%, StanChart and CBA are around -5% net — a “gradual restructuring” rather than “apocalypse” pattern.

Altman’s direct quote that he “really value the relationship with people” signals domains AI cannot fully replace short-term clearly exist. Premium sales·B2B, healthcare·counseling·psychology, education·coaching, leadership·negotiation, and field service·care are the 5 core domains.

These share one trait — non-verbal elements like empathy, trust, contextual understanding, and physical contact are central. Text/voice AI interfaces cannot fill the “human value” decisive in decisions and customer satisfaction. These industries may be re-rated as defensive plays in the AI era.

While Altman’s message is “not apocalypse”, roles actually under threat clearly exist. Call center/tier-1 support, code writing/debugging, routine accounting/reports, content/image/design, translation/summary/document processing see the most direct impact.

Call center and tier-1 customer support are automating fastest. Simple inquiries/FAQs are 90%+ handled by AI chatbots, with human agents specializing into complex/VIP cases. Code writing also shows multiple cases where developer productivity rose 2~3x using GitHub Copilot, Cursor, etc.

Combining McKinsey/MIT Tech Review analyses with Altman’s “gradual” tone, white-collar role change rates over 2026~2031 split into 3 scenarios. Bull 27% / Base 45% / Bear 70%. Bull aligns with Altman’s corrected view; Bear matches his earlier “apocalypse” prediction.

Markets and policymakers currently see the Base scenario (45%) as most likely. I.e., ~45% of white-collar workers will need to actively use AI by 2031, with major changes in role content and required skills. This creates retraining/transition pressure for workers, AI investment pressure for companies, and labor policy/retraining infrastructure pressure for governments.

Altman’s “gradual” message sends two signals to investors. First, AI infrastructure firms (NVDA·MSFT·Palantir·Oracle etc.) see higher long-term demand stability. If AI rolls out over 5~10 years gradually rather than exploding, infra revenue follows a steady up-trend with lower short-term volatility.

Second, human-relationship industries (healthcare·education·premium services) get re-rated as defensives. UNH (UnitedHealth), JNJ (Johnson & Johnson), LVMH (luxury), Accenture (consulting) receive safety premium as “AI-resistant”. Balanced portfolios with “AI infra + human-relationship defensives” make sense.

Five investment strategies based on Altman’s “AI jobs apocalypse X”: NVDA·MSFT phased buying + Palantir + healthcare defensives + Korean HBM + risk hedge. Balance AI infra and human-relationship defensives as two axes; cap combined exposure at 30~40%.

#SamAltman
#OpenAI
#ChatGPT
#AIJobs
#AIJobsApocalypse
#AIUnemployment
#WhiteCollar
#HumanRelationships
#AILimits
#Nvidia
#NVDA
#Microsoft
#MSFT
#Palantir
#PLTR
#HealthcareDefensives
#UNH
#JNJ
#LVMH
#Accenture
#SamsungHBM
#SKHynix
#AIInfrastructure
#GradualAdoption
#Retraining
#RoleTransition
#getdirEconomicInsights
#getdirAltman
#jyfamilyoffice
#2026AIInvesting