AI Mania Is Eviscerating Global Decision‑Making – A Consultant’s Front‑Row Account

AI Mania Is Eviscerating Global Decision‑Making – A Consultant’s Front‑Row Account

TL;DR – AI hype is killing decision‑making

The author’s front‑row experience with dozens of AI projects across Fortune‑500 firms shows a 0 % success rate and a culture of forced AI adoption that rewards hype over results. Executives, fearing career death, double‑down on AI even when internal data shows no usage or impact. The result is a coordination problem where honest dissent is punished, leading to wasted budgets, toxic politics, and a halt to sensible work.

1. AI Investments Are Mostly Failures

  • Observed failure rate: The author’s team has seen zero successful AI projects in a year‑and‑a‑half of work, including projects they were not directly hired for.
  • Why projects fail:
    1. Execution problems – most companies are bad at running software projects; adding LLMs only compounds existing failure modes.
    2. Technology limits – internal chatbots never see adoption because documentation is poor and LLMs are not “psychic.”
    3. Metric gaming – leaders avoid tracking real usage, instead counting licenses or token consumption, which can be inflated without delivering value.
  • Anecdotal evidence: A Mitsubishi customer‑service bot promised a callback that never arrived; the author never learned whether the request was logged as resolved.
  • Commentary: Several commenters note the author’s claim of 0 % success may be hyperbolic, but they agree that most high‑profile AI roll‑outs deliver little measurable productivity and often hide behind vague definitions of “AI.”

2. Speaking Against AI Is Career Suicide

  • Cult‑like pressure: In firms with >500 employees, executives repeatedly proclaim “AI is changing everything” even when they have never used an LLM themselves.
  • Punishment for dissent: Employees who question AI strategy are often fired or sidelined; the author cites examples of high‑performers being let go for achieving results without AI.
  • Token‑leaderboards: Some teams are judged on raw token usage, incentivising superficial AI adoption (e.g., rewriting code in a new language just to log AI usage).
  • Commentary: HN users echo this, calling the environment a “cult” and noting that dissenters are “gunned down” in the proverbial streets of corporate politics.

3. Flashy Demos Undermine Rational Decision‑Making

  • Snowflake Cortex demo: A live demo of a natural‑language query engine convinced skeptical prospects to buy immediately, despite the author’s warning that accuracy is ~92 % and unsuitable for production.
  • Result: The sales team refused to close the deal to avoid reputational risk, illustrating how demo‑driven hype can override sound business analysis.
  • Commentary: Multiple commenters point out that such demos are “mind‑killers” that create a false sense of progress, similar to past tech manias (e.g., blockchain).

4. Executives Face a Prisoner‑Dilemma

  • Game‑theoretic trap: Executives must either co‑operate (admit AI projects are failing) and risk being fired by peers, or defect (double‑down on hype) and risk being replaced by a more compliant colleague.
  • Board pressure: Board members admit skepticism but feel their jobs depend on visible AI investment, leading to a coordination problem that stalls honest dialogue.
  • Commentary: Several HN participants note this mirrors classic “emperor’s new clothes” scenarios where everyone pretends success to avoid being the outlier.

5. The “AI‑Native” Requirement Stifles Real Work

  • AI‑first mandates: Projects are re‑scoped to include any AI component, even when it adds no value (e.g., a database migration billed as “AI‑driven” because a few lines of SQL were generated by an LLM).
  • Metrics become meaningless: Employees are judged on AI spend or token counts; refusing to add AI can lead to being labeled “bad at AI” and terminated.
  • Commentary: Critics argue the author’s selection bias (consulting firm that only takes on struggling projects) inflates the failure narrative, but they agree that AI‑label inflation is widespread.

6. How to Survive the AI Mania

6.1 When You Have Other Objectives

  1. Private dissent: Raise concerns one‑on‑one, not in group settings, to avoid peer retaliation.
  2. Anonymous polling: Use surveys to surface a bimodal view of project health (e.g., many rate a project 3/10 while a few claim 8/10).
  3. Ground‑level data: Gather feedback from actual users; many staff are unaware they even have AI licenses.
  4. Avoid direct challenges: Let statements like “AI is changing everything” slide unless you need to influence strategy at the highest level.
  5. Strategic lying: In extreme cases, add a token‑heavy AI component to keep your job, but plan an exit strategy.

6.2 When You Just Want to Stay Sane

  1. Accept limited influence: Recognize that you may not be able to change the AI narrative; focus on personal well‑being.
  2. Contracting: Consider short‑term contracts to escape toxic politics and retain higher pay.
  3. Information diet: Limit exposure to AI hype (avoid HN, Reddit, etc.) to preserve mental health.
  4. Deflect pressure: When asked for AI opinions, downplay its impact and steer conversation elsewhere.
  5. Prepare to leave: Treat every large AI request as a signal to start a job search; the burnout is often inevitable.

7. Community Reflections

  • Agreement on hype: Many commenters (e.g., @hliyan, @simonw) confirm that internal chatbots see little use and that executives often lack personal AI experience.
  • Critique of hyperbole: Some (e.g., @A1kmm, @azakai) argue the author’s blanket “0 % success” claim is exaggerated and note successful niche uses (semantic search, code generation).
  • Selection bias concern: @kbar13 and @Aurornis point out the author’s consulting focus may skew the sample toward failing projects.
  • Broader analogy: @zkmon likens AI mania to historic tech fads (tulip mania, agile, timesheets), suggesting such cycles are inevitable.
  • Technical nuance: @jdw64 notes the post lacks concrete technical analysis (e.g., RAG, vector‑DB quality), but agrees the overall message about over‑hyped expectations holds.

8. Final Thought

The AI frenzy has created a self‑reinforcing feedback loop where executives must appear AI‑savvy, employees must fabricate AI usage, and real productivity is sacrificed. While the hype will eventually subside, the damage to decision‑making processes and corporate culture is already deep. Professionals can mitigate personal risk by quietly gathering data, avoiding public dissent, and planning exits, while organizations that break the cycle—by measuring real outcomes and allowing honest discourse—will be the true survivors of the AI mania.

Sources