You Are The Line Item

AI Is Not A Transition

AI is going to cost me my job in the future, isn’t it? The answer is yes, there is a fair chance it will. This paper is not intended to include a sprinkle of sugar. It is intended to be an echo of warnings that have been given by many over the past few years and are getting more and more frequent for a reason.

Calling the current AI shift a transition obscures reality. Companies deploy AI to cut costs and scale output. The law rewards that behavior, not job preservation.

Humanoid robots are likely 6-8 years away and combined with Artificial General Intelligence (AGI) make an employee very inexpensive, pricing humans out of the workforce in jobs that can be automated and most jobs with a computer.Corporate leadership must try to make money for shareholders by law. For the non-technical, AGI and/or Super-intelligence can be roughly described as when the system is smarter than us humans, predicting anything we may think or do.

AI-driven efficiency will collapse the labor denominator in most white-collar and administrative sectors. There is currently no credible market mechanism to reabsorb displaced workers at scale within five years or even eight years, hence the continued disbelief by many of the potential impact on their lives. Experts like Roman Yampolskiy, Geoffrey Hinton, and Eliezer Yudkowsky have been warning for years now of the potential risks to society, jobs, etc.

Person	Public risk signal	Numeric (if stated)
Roman Yampolskiy	“Nearly certain” loss-of-control framing	“Nearly certain” Business Insider
Eliezer Yudkowsky	“Near-100%” doom	~100% (as written) LessWrong
Dan Hendrycks	High doom probability	>80% X (formerly Twitter)
Toby Ord	AI as a major x-risk driver	~10% AI x-risk (100 yrs) Effective Altruism Forum+1
Geoffrey Hinton	Explicit extinction odds	10-20% (30 yrs) The Guardian
Max Tegmark	Cites serious extinction odds among researchers	cites ≥10% as common; also discusses high personal concern TIME+1

tight p(doom) / systemic-failure table (where people give numbers)

Let us look at some harder topics and numbers now. On the “AI is going to cost me my job, isn’t it?” question, we have to look at the most exposed job categories. Below is a sample of roles easily replaced by AI:

• Administrative and clerical work
• Office support roles
• Finance and banking operations
• Real estate brokerage
• Customer service and call centers
• Basic legal and compliance work
• Marketing and copy production
• Transportation and driving (6-8 year horizon)
• Middle management coordination
• Routine analytics and reporting

I will try and make this relatable: this equates to 2 million jobs in 5 years in the US (+/-2million). This is conservative relative to McKinsey and Goldman Sachs. Taking a look elsewhere, in the Philippines, call center jobs account for about 1.7 million jobs +out of a total workforce of 49 million (3.7%). In India where the services sector accounts for around 188 million jobs it is apx. 31% of all jobs (nearly 1/3). By the 5 year mark, Philippines call center/BPO proxy job loss to AI would be 270k-640k jobs. In India, losses in services to AI would be between 8 and 18% (similar 5 year period) which translates to 15m-34M jobs. This services loss would be front loaded in retail/admin/call-center then slower uptake in care and field services. As time rolls along, by year 8 it does not get better.

Overall Global Numbers

3-5 year low-high AI displacement ranges:

Services (~1.85B):

• Office & admin (~400M): Low 30% / High 60% → 120-240M
• Retail & sales (~550M): Low 15% / High 35% → 80-190M
• Hospitality & food (~350M): Low 10% / High 25% → 35-90M
• Healthcare & care (~300M): Low 5% / High 15% → 15-45M

Industry (~850M):

• Manufacturing (~350M): Low 15% / High 35% → 50-120M
• Construction & trades (~250M): Low 5% / High 15% → 12-40M
• Transport & logistics (~250M): Low 10% / High 30% → 25-75M

Agriculture (~750M):

• Smallholder & ag labor (~650M): Low 5% / High 15% → 30-100M

6-8 year Revised ranges :

• Office & admin (~400M): Low 40% / High 75% → 160-300M
• Retail & sales (~550M): Low 25% / High 50% → 140-275M
• Healthcare & care (~300M): Low 10% / High 30% → 30-90M
• Transport & logistics (~250M): Low 25% / High 60% → 60-150M

Assessment by category with comments on estimate factors:

• Office & admin (40-75%) – correct; 75% is believable once agentic workflows and manager substitution mature.
• Retail & sales (25-50%) – solid; physical retail friction keeps it below admin.
• Healthcare (10-30%) – upper bound is aggressive but valid if scoped to admin, triage, documentation, and monitoring, not bedside care.
• Transport & logistics (25-60%) – correct and possibly understated at 6-8 years if autonomy clears regulatory drag.

(These subdivisions use logical distribution within sectors, not a single dataset – but reflect common patterns from ILO/FAOSTAT employment trends.)

Sources:

1. World Bank World Development Indicators (Employment by sector) – modeled ILO estimates for services, industry, and agriculture percentages.
   Search: Employment in agriculture (% of total employment), Employment in industry (%), Employment in services (%) in World Bank DataBank. https://databank.worldbank.org/id/ebc357b9?utm_source=chatgpt.com

1. Our World in Data on employment by sector (ILO data) – shows services as ~50 % and agriculture ~27 % globally. https://ourworldindata.org/data-insights/the-service-sector-now-represents-about-half-of-employment-across-the-world
2. FAOSTAT Employment Indicators – gives latest numbers for agriculture employment (~892 M). https://www.fao.org/statistics/highlights-archive/highlights-detail/employment-indicators-2000-2023-(july-2025-update)/

This matters because large-scale displacement shifts costs, not value, onto welfare systems.

The million-dollar question: What jobs are safe? Trades, crafts, arts, outdoor work, and human-authentic services gain premium value.

By now, you are likely asking yourself, if this is true, and the experts know, why aren’t we doing something about it? The short answer is it’s a narrative failure. Expert commentary now acknowledges job loss but re-frames it as gradual and absorbable. This framing is unsupported by revenue economics and labor demand trends. Political incentives further delay action, increasing systemic shock severity.

What can we do? We can ease the impact of AI. We cannot reverse it.

If displacement is inevitable, policy can only slow the rate of damage and shape who absorbs the shock.

Mitigation Actions

• Annual national caps on AI-driven job replacement by industry/year.
• Prohibition on AI deployment tied to net staff reductions or reclassified replacements.
• Human-staffed government oversight offices to audit AI versus human labor displacement.
• Explicit closure of reclassification and offshoring loopholes.

Adaptation Imperative

• Early reduction of welfare dependency combined with aggressive skill diversification is required to push populations toward secure trades, crafts, and human-authentic work.

*AI assisted in drafting and imagery; all analysis and decisions remain human.