IRS lost 40% of IT staff, 80% of tech leaders in 'efficiency' shakeup

AI at the IRS and “AI Miracle” Culture

• Several commenters are alarmed that the IRS plans to use AI/LLMs on financial data where accuracy is “table stakes,” and worry about privacy and hallucinations.
• Others note that AI could hand off precise math to traditional systems and be used mainly for pattern‑spotting (“does this return look legit?”).
• Treasury docs cited in the thread indicate they do mean LLM-style tools (AI chat, AI-assisted coding), not narrow finance models.
• Broader frustration: managers in many orgs are demanding an “AI miracle,” ignoring staff warnings and degrading UX with bots and call-center AIs.

Defunding, Politics, and ‘Starve the Beast’

• A strong current argues the cuts are part of a long-running strategy to weaken the IRS so complex tax evasion by the rich goes unenforced.
• Two motives are described: rich taxpayers using complexity as a shield, and ideologues who want to shrink federal government by cutting its revenue and then pointing to dysfunction.
• Others push back on blanket claims about one party’s motives, calling them hyperbolic, but supporters respond with examples like appointing tax cheats and clawing back IRS expansions.

Who Gets Audited: Poor vs Wealthy

• Data shared: nearly half of 2022 audits were on filers under $25k with EITC; 87% were on people under $200k, suggesting current practice targets the poor more than “uber rich.”
• One side argues these low‑income audits are often automated, low-penalty corrections (missed W‑2, misclaimed credits).
• Others counter that gutting the IRS will only further reduce capacity to pursue complex high‑net‑worth cases, which was the purpose of recent staffing increases.

IRS Funding, ROI, and Hidden Costs

• Multiple comments cite very high returns on IRS funding (numbers from ~10:1 up to 415:1), used to argue the IRS is underfunded and extremely efficient at raising revenue.
• Former federal audit experience is invoked to say 415:1 is misleading; agencies typically target ~10:1 because returns fall off and compliance/indirect costs balloon beyond that.
• There’s agreement that additional enforcement has diminishing returns and significant second‑order economic costs, but also that current funding is far from the point of over-enforcement.

Size and Role of IRS IT; Impact of Cuts

• Some see 8,500 IT staff as “insane” for an agency that outsources a lot and has few visible products.
• Others note the IRS serves ~150M individual filers, multiple digital services (including Direct File), and still trails countries like the UK in IT staff per capita.
• Examples of creaky processes (EIN via fax after online failure) are used to argue the IRS does not, in practice, have “too many” tech people.
• Reported numbers show about a 16% IT headcount reduction year-over-year, raising questions about how “40%” was calculated and whether this is a rollback of a recent hiring surge or deeper hollowing-out.

Workforce Quality, Layoffs, and Morale

• One line of commentary assumes significant deadweight in public-sector roles; others strongly reject this, saying most government workers take pride in their jobs.
• Several argue that layoffs rarely remove the “bottom 10%” but often shed good people who want out, whole teams deemed expendable, or those caught in politics, with capability loss roughly proportional to headcount lost.
• Some in private industry say AI mandates and layoff threats are making them want to quit, suggesting similar morale risks inside the IRS.

International and Structural Comparisons

• Non-US commenters note that in some countries (e.g., Argentina), the government simply tells citizens what they owe; no annual filing is needed.
• Multiple participants argue the core US problem isn’t IRS headcount but a politically maintained, lobbyist-influenced tax complexity that both burdens ordinary filers and enables sophisticated avoidance.