Supporters cast effective altruism (EA) as the ultimate rational framework for improving human welfare, while critics call it cold and callous, with the truth sitting somewhere in the probabilistic middle. By treating philanthropy as a global optimization problem driven by spreadsheets, linear regression, and cost-benefit calculations, EA earned a reputation as the young Sheldons of the charity world.
I’ve deliberately stayed out of EA, with no donations, no events attended, and no community participation, because to me the worldview feels unhealthy, it feels like a death cult. From that considered outsider distance, we can test which criticisms survive when measured against EA’s strongest stated arguments.
Key Takeaways
Effective altruism began as an evidence-based effort to route money away from zero-impact programs toward verified wins like malaria nets and GiveDirectly cash transfers, anchored by Peter Singer’s 1972 pond thought experiment and Giving What We Can’s 2009 10% lifetime income pledge.
Strict cost-effectiveness math using quality-adjusted life years (QALYs) pushes moral arithmetic to strange extremes, from the Shrimp Welfare Project claiming $1 improves conditions for over 2,000 shrimp to Gates Foundation-sponsored Kurzgesagt videos favoring status-quo “vote with your wallet” fixes over unmeasurable systemic tax and economic reform.
Backed predominantly by Silicon Valley tech wealth, institutional EA drifted from global poverty toward speculative artificial general intelligence (AGI) extinction risk for trillions of nonexistent future humans while suffering a defining reputational blow when major funder Sam Bankman-Fried’s FTX collapsed in late 2022.
Table of contents
Table of Contents
What is effective altruism, and how is it different from ordinary altruism?
Effective altruism is a philosophical and social movement that differs from ordinary altruism by treating ethics as a quantifiable cost-benefit problem rather than relying on emotional impulses or local loyalties. While consensus within the movement holds that applying scientific-style evidence and careful reasoning reveals the most effective ways to improve the world, that calculation demands routing money and careers strictly where a dollar buys the largest measurable gain. Two core mechanisms define the architecture:
- Philosophical impartiality from Peter Singer’s 1972 paper Famine, Affluence, and Morality: $100 means little to an affluent Briton or American, yet that same sum funds disaster relief for distant victims of a Bengali flood.
- Institutionalized optimization through Oxford-founded groups like Giving What We Can (2009), where members pledge at least 10% of their lifetime income to top-ranked charities, and 80,000 Hours (2011), which applies QALY cost-effectiveness math to career choices.
Peter Singer’s drowning child and the demandingness bug
In 1972, Singer posed a thought experiment: if you pass a shallow pond on your way to give a presentation and spot a child drowning, ruining an expensive suit and missing the event are nothing next to saving the child’s life. Because geographical distance is no ethical excuse not to help, Singer argued affluent people have the same moral imperative to fund distant disaster relief. Rooted in the utilitarianism of Jeremy Bentham and John Stuart Mill, which seeks the greatest happiness for the greatest number, EA assumes the good is measurable and popularized that view in Singer’s 2015 book The Most Good You Can Do.
Run that code to completion, however, and you hit the demandingness problem: even money spent on a cinema ticket ought to be donated, and Singer’s book opens with a figure named Matt who donates blood, bone marrow, and a kidney to a stranger. Students taught Singer’s paper typically feel a strong pull toward the conclusion in the classroom, then quietly renegotiate it afterward in practice.
The Oxford founding and the 10% compromise
Ten to 20 years ago, two Oxford philosophers influenced by Singer, William MacAskill and Toby Ord, decided philosophy should change the world rather than just understand it. Traditional charity budgets often went to wasteful programs, unlike animal shelters saving very few lives per dollar compared to malaria prevention, or an African school program explained in a movement post-mortem:
And it just turned out that the measured impact of the better textbooks was like zero, Right? And so effective altruism comes on the scene and it’s like,
When MacAskill and Ord founded Giving What We Can in 2009 with its 10% lifetime income pledge, they created a livable compromise, even as MacAskill maintained that billionaires ought to donate 99% of their income.
The measurement machinery: QALYs, guide dogs, and the shrimp problem
Measuring altruism strictly by cost-effectiveness fits donors comparing well-defined medical interventions inside a trillion-dollar global charity industry where over 3 billion people a year donate time and money, but it breaks down the moment human ethics involves values you can’t put in a spreadsheet. To rank charities, EA borrowed the QALY from economists.
Per MacAskill, stroke survivors rate their life quality at roughly 75% of prior quality. Training one guide dog costs roughly $50,000, whereas that same spend treats or prevents blindness for many more people. In Doing Good Better, MacAskill compared that price gap to buying a round of drinks at $1,000 each versus a penny each. Yet in a Doom Debates opening statement, sociologist Jonas Söderberg argued that maximizing impact fails because ethics involves multiple incommensurable values: you can’t do ethics like math.
When you treat all welfare as a single continuous scale differing only in quantity, the math goes weird fast. Former investment banker Andrés Jimenez Zorrilla founded the EA-funded Shrimp Welfare Project after witnessing the refugee relief work his wife was doing. A Daily Show segment with Ronny Chieng showed the group fighting shrimp suffering at slaughter via electrical stunning, avoiding overcrowding, and ending eyestalk ablation (cutting or crushing female shrimp’s eyes). With 50 million shrimp harvested per hour on farms, the group claims $1 improves conditions for over 2,000 shrimp and that 2.7 billion shrimp received improved conditions in the prior year.
EA defenders make a cost-effectiveness argument for shrimp aid: a few dollars can improve conditions for thousands of shrimp, so if shrimp suffering is even a tiny fraction, at thresholds of one-thousandth to one-in-a-million, as intense as human suffering, helping vast numbers of shrimp justifies some shrimp aid, even if the debating defender had not personally donated while believing the cause deserves money. On The Politics of AI podcast by the Catholic channel New Polity, speakers Alex Denley and Mark Barnes (associated with the College of St. Joseph the Worker) recounted a convert’s testimony about an EA member who calculated the suffering of ordering shrimp at restaurants and donated to an EA-approved charity to offset it, like a moral carbon tax or reinvented Catholic indulgences.
Denley and Barnes argue EA’s metric isn’t fundamentally human, contending that most followers believe a large enough quantity of shrimp outweighs a drowning child or 10,000 Bengali famine victims (a dilemma another commentator joked you could solve if you feed the Bengalis the shrimp). Philosopher Elizabeth Anscombe had little patience for these Oxford-style hypotheticals. Asked whether she would kill 20 good-looking gentlemen with her left foot or 50 horrible fat people with her right foot, she famously replied that she would jump using both feet at once.
Why effective altruism turned from malaria nets to AI doom
Silicon Valley tech-billionaire funding and Oxford’s “longtermist” math pushed effective altruism away from global poverty relief toward speculative AI doomerism, weighting trillions of hypothetical future humans above people suffering today. Before its rapid financial expansion from the mid-2000s onward, global poverty and health were EA’s primary topics. The diff between William MacAskill’s two flagship books proves how sharply the movement’s priorities shifted alongside its tech-industry bankroll:
- Doing Good Better focused overwhelmingly on global poverty and health, mentioning existential risks only at the very end and far less prominently.
- What We Owe the Future extended moral concern to trillions of future, nonexistent people, backing an institutional pivot that now spends millions of dollars on existential risk theory and AGI research on the bet that superintelligent AI could erase those future lives, a turn over which Steven Pinker publicly expressed disappointment, and which critics correlate directly with Silicon Valley billionaire funders.
| Era and flagship text | Primary moral target | Core funded interventions | Existential risk weight |
|---|---|---|---|
| Early EA (Doing Good Better) | Present global poor and sick | Malaria nets, blindness treatment, cash transfers | Partial (brief mention at very end) |
| Longtermist EA (What We Owe the Future) | Trillions of nonexistent future people | AGI alignment, existential risk theory | Yes (millions spent as top priority) |
From Oxford circle to tech-funded movement
What began in the mid-2000s as a small circle of philosophers and activists grew into a massive apparatus centered on Oxford University and the Centre for Effective Altruism, which runs Effective Altruism Global. At Oxford, Nick Bostrom was a leading figure whose Future of Humanity Institute (FHI) influenced the 2011 creation of 80,000 Hours. Institutional milestones piled up quickly: in 2007, former hedge fund employees Holden Karnofsky and Ellie Hassenfeld founded an EA-aligned charity evaluator, and Facebook co-founder Dustin Moskovitz teamed up with Carrie Tuna to establish Open Philanthropy.
The longtermist pivot
Here at GeekExtreme, we’re used to watching a clean v1.0 tool bloat into an unrecognizable enterprise suite once venture money arrives, and EA’s longtermist turn follows that exact pattern. As an EA-community blog post by a writer named Finn outlines, the longtermist thesis depends on three premises: future people matter morally, the future could contain vastly more people than have lived so far, and present actions shape their well-being across thousands of years.
Critics in The Rise and Fall of Effective Altruism analysis point out that gambling millions on speculative AI doom is the exact low-probability bet EA normally rejects. They conjecture that Silicon Valley billionaire funders, with figures like Elon Musk in the wider orbit, steered institutional EA toward a savior complex that ignores present-day harms such as algorithmic bias and the scraping of artists’ work for AI training.
The pipeline into frontier AI labs
That Oxford academic pipeline feeds directly into frontier AI labs like OpenAI and Anthropic. Joseph Carlsmith, a leading alignment researcher at Anthropic, previously worked for an EA nonprofit and served as a research fellow for Toby Ord’s book The Precipice.
On the New Polity podcast, commentators identified Carlsmith and Anthropic colleague Amanda Askell as EA members, noted that CEO Dario Amodei has hovered around EA crowds, and cited a report that Anthropic was preparing for an IPO at an expected $2 trillion valuation. Meanwhile, the broader AI field has fractured between EA-aligned safety doomers, who debate internally whether superintelligence will kill everyone and coordinated an international Pause AI protest on February 7 to 11 alongside the Paris AI action summit, and a national-security camp led by Alexander Karp and Peter Thiel.
That right-wing camp in power holds that AI’s purpose is national supremacy and pressures EA-influenced developers to reframe their work around national security. As the New Polity critics observe, adopting nationalism isn’t foreign to EA logic: earn-to-give already blesses doing whatever current power structures require to accumulate wealth and influence for later good.
When funding shapes the message: Kurzgesagt and enlightened centrism
In the second half of 2022, several YouTube videos criticized the explainer channel Kurzgesagt because its climate, overpopulation, parasites, and vaccine videos were all sponsored by the Gates Foundation, it prominently featured a MacAskill longtermism book, and Open Philanthropy helped finance its short-form videos and translated editions, several of which touched on EA themes. The channel’s sponsored climate episode urged viewers to “vote with your wallet” via electric cars, renewable energy, and carbon capture, which critics called textbook enlightened centrism. Philosophy Tube argued that while EA is theoretically neutral between revolutionary, centrist, and conservative paths, its quantification bias filters out systemic reform: you can calculate banking or bed-net returns, whereas restructuring economic or tax systems can’t be cleanly quantified.
Earning to give, televangelists, and the Sam Bankman-Fried collapse
Yes, the FTX collapse and Sam Bankman-Fried damaged effective altruism because he was one of the movement’s largest funders and the public showcase for its “earning to give” doctrine until his crypto exchange’s late-2022 bankruptcy turned EA into a mainstream cautionary tale. Promoted by 80,000 Hours (founded in 2011 and named for the typical working lifetime), earning to give argues that taking a lucrative finance or corporate job and donating heavily beats direct service work like medicine or social work.
When FTX imploded in late 2022, institutional EA went quiet and William MacAskill stepped back to fundraising and background projects. Media coverage receded into post-mortems, including a Scott Alexander blog post summarizing EA’s achievements and AI safety work, while one EA debater insisted Sam Bankman-Fried was a random example rather than the epitome of the movement.
Beyond crypto fraud, the earn-to-give career path faces a deeper structural problem. Critics on the New Polity podcast argue that the pipeline is self-defeating: because charity and love are infectious dispositions learned from other people’s example, routing idealistic young people through an individualistic, profit-maximizing corporate environment predictably produces competitive, egotistical people and erodes the very altruism it presupposes through career escalation and deferred giving. Those Christian critics point out that earn-to-give mirrors how televangelists justify wealth, letting wealth-builders cast themselves as heroes in the present by pointing to a promised future good that nobody audits today.
Who defines the good: TOMS, PlayPumps, Kenya, and San Francisco
When distant donors optimize spreadsheet metrics without local skin in the game, two questions expose the power imbalance: who defines what counts as good, and why does a movement concentrated in the San Francisco Bay Area leave its own streets in misery?
When distant giving backfires
Distant charity fails like TOMS Shoes or PlayPumps International whenever well-intentioned funders air-drop measurable outputs that ignore local economies and institutions, though EA defenders counter that such flops are “ineffective altruism” by definition. TOMS Shoes’ buy-a-pair-donate-a-pair model flooded communities with free shoes and put local cobblers out of business, while PlayPumps’ carousel water pumps proved wasteful, barely used, and net-negative.
In practice, wealthy donors decide what counts as good rather than recipients: New Polity speakers noted that American billionaires funded contraception promotion in Kenya despite opposition from Kenyan Catholics, while a Doom Debates guest cited Gaza aid as a live test case where donor metrics diverge from recipient realities, noting critics’ concern that donors may be unaware aid reaches Hamas and increases its leverage (though Gaza barely registers on the EA forum’s cause list). In a donations-versus-development debate, the interviewer cited African leaders who warn that Western aid keeps Africa dependent. The EA-aligned interviewee conceded that air-dropping free shoes wrecks local producers while arguing that most donated cash is spent locally, and the interviewer countered that cash injections are small compared with building trustworthy local institutions and stable governance, like the Andrew Carnegie library in Steubenville, Ohio, built when wealthy elites felt a domestic obligation rather than practicing distant “telescopic philanthropy.”
The local problem EA’s own backyard can’t solve
Whether effective altruism ignores local problems like San Francisco homelessness comes down to a clash between EA’s roughly 80% foreign / 20% local focus ratio and defenders’ claim that such misery is a governance failure where individual donations barely move the needle. The San Francisco Bay Area has the highest concentration of EA members per capita, yet in the debate, 100% of surveyed acquaintances said they wouldn’t walk alone on San Francisco streets at night, and the city reportedly spends $80,000 to $90,000 per homeless person annually on what critics call a social problem rather than a funding gap.
Critic Jonas Söderberg proposed flipping EA’s 80/20 ratio over a 5-, 10-, or 20-year horizon by training young American leaders in their mid-20s to early 30s with cross-cultural understanding to guide US policy. EA’s defender countered that fixing a functioning US society models success globally, and that EA cause areas include founding new organizations, just as EA members founded Africa-focused charities when existing ones proved inefficient.
The steelman: what effective altruism’s defenders get right
When critics claim effective altruism is bad, defenders reply that the entire point of EA is doing basic math to avoid misguided charity, separating verified high-impact interventions from flops like PlayPumps. At GeekExtreme, we don’t call a system broken without checking where its code actually executes cleanly, and EA’s defenders bring real empirical receipts.
In the hosted debates, the EA-defending host coined the term “recoil exaggeration” for critics’ bias of inflating unintended side effects until they seem to cancel out the forward benefit. He defended malaria nets as a near-uncontroversial win with roughly a 100x impact multiplier over typical giving, dismissing Marc Andreessen’s tweeted “study” claiming nets get used for fishing as non-credible character assassination while noting the irony that Andreessen defended Mark Zuckerberg’s hospital charity the exact same week. Similarly, GiveDirectly sends no-strings-attached cash transfers: a donor-speaker in the debate gives recurring ~$50 monthly donations (often modeled around a $1,000 annual donation benchmark) to people in Africa living on ~$2 a day, where ~$50 buys immediate breathing space like a waterproof roof that frees up time to run a business.
Both sides in the development debate converged that over a decade-long or 100-year timeline, making capitalism work better beats one-time donations, citing Jeff Bezos’s final Amazon shareholder letter on consumer surplus and a year-1600 thought experiment where heavy charity without capitalism yields a 2025 with no computers or internet. The EA interviewee agreed governance work could beat a mosquito net long-term even if he hasn’t seen EA prioritize it, though defenders note you can’t donate to capitalism and people today still need bed nets.
On partiality, Sam Harris argued in a Lex Clips segment that prioritizing one’s nearest and dearest can be ethically optimal inside fair systems: he wants a hospital that doesn’t treat his sick daughter better than other kids even though he worries about her most, noting that fairness norms erode in extreme lifeboat emergencies that shouldn’t model normal life. The debate host, who ranks Doing Good Better among his top 10 books read, argued EA’s logic would be reinvented if the movement died, while EA blogger Finn conceded most members still genuinely want to do good better despite the “AI takeover,” advising readers to use numbers without getting lost in them, skip the sales pitch, and stop putting faith in crypto billionaires who advertise their own frugality.
Is effective altruism a death cult? The culture question
Effective altruism isn’t a literal religious death cult, but its hyper-impartial utilitarian culture earns that visceral outsider reaction whenever its spreadsheet optimization trains adherents to override ordinary human attachment. You can feel that friction in the room: in the Doom Debates exchange, critic Jonas Söderberg reported from sociological observation that EA community events don’t feel like wholesome places because they suffer from a lack of emotional attunement, a charge the EA-defending host half-conceded from personal guilt. As the Daily Show explainer noted, while traditional charities like the ASPCA lean on emotional Sarah McLachlan commercials, EA gets trapped in recursive self-optimization, asking what is the most effective way to do effective altruism itself, which drove post-FTX effective altruists and accelerationists into fretting over existential risks while bloggers like Bentham’s Bulldog debate extremes on the EA forum.
That cultural disconnect crystallizes in the burning-building dilemma cited in The Rise and Fall of Effective Altruism: if a burning building contains both a child and a priceless Picasso, some contend that rescuing the canvas is the rational choice because its sale could finance life-saving programs with a higher total headcount. Set that beside trading off shrimp welfare against a drowning child or placing trillions of hypothetical future humans first, and you see the exact conditioning that triggers my “death cult” intuition.
The Catholic counter-framework: what the theological critique sees that secular critics miss
Catholic social teaching recommends choosing subsidiarity, loving what is near, and Pope Paul VI’s integral human development over effective altruism’s utilitarian charity, favoring local accountability, distributed property ownership, and state-level obligations to train developing nations’ students and fight corruption rather than relying on distant NGO wealth transfers. That framework, advanced by the New Polity school and papal encyclicals, challenges effective altruism at its root by arguing that utilitarianism cannot define “the good” on its own, functioning instead as an individualistic, technocratic ideology that feeds on pre-existing Christian generosity while bypassing local accountability and state-level development.
EA as parasitic on pre-existing charity
Launched to examine the philosophical, theological, and anthropological beliefs built into AI systems, drawing on thinkers such as René Girard after the Pope urged discernment over what guides AI design. New Polity’s The Politics of AI podcast contends that EA is both asocial and parasitic. Because the organizations EA evaluates are not themselves driven by EA, the movement relies on non-EA charities existing and consumes a sacrificial-giving disposition that the speakers attribute to Christianity’s legacy in the West. Against consequentialism, they cite St. Paul’s rule that you cannot do evil so good may come, contrasting EA’s aim to do all the good one can with wealth against the rich young man commanded to give everything away and follow Christ, who walks away sad because of his possessions.
While Singer’s call to live modestly parallels Lenten teaching and Christian charity is obligatory regardless of feeling (unlike subjective empathy exercises in traditions like the Dalai Lama’s), the hosts observe that a Catholic approach to curing blindness would first consider patient consent and the meaning of redemptive suffering, whereas EA takes the definition of the good as self-evident, as seen in the Kenya contraception campaigns. They call EA “gasoline for the fire” of tech development that excuses present harms for future good, sharing a technocratic paradigm with transhumanism that brackets immediate human happiness for quantitative outputs. By contrast, the Pope in Magnificat Humanitatis states technology is never neutral because it reflects its designers’ morality, rejects transhumanism, and posted that there is an ontological, before an aesthetic, difference between human work and AI output.
The Catholic alternative: subsidiarity and integral human development
Politically, the New Polity speakers argue EA acts as an ideological defense of tech oligarchy, with Bill Gates as the exemplar. The movement treats wealthy individuals and NGOs as the only effective agents of good and presumes states cannot achieve mutual development. Against Singer’s global $100 calculus, Catholic social teaching pairs subsidiarity, loving according to what is near so recipients can hold donors accountable in their own language and presence, with Pope Paul VI’s “integral human development” (reiterated by Pope Leo in the encyclical Dilexi Te).
Integral human development encompasses the whole person, respect for local culture, and civic participation. Wealthier nations, rather than billionaires, carry the obligation to train developing countries’ best students, share expertise, fight corruption, and build ownable investment. Distributed ownership, liberty through property so families and nations become self-sufficient, replaces GDP as the success metric, making a global NGO in a world of self-sufficient nations look like a tyranny and foreign invader.
The speakers argue the Catholic Church has done more for developing-world development than any state or group via local schools, hospitals, and parishes, citing Ivan Illich’s Mexico City speech telling UN Peace Corps volunteers to “go home.” Reporting conversions to Catholicism including a former EA member, they recommend Benedict XVI’s Deus Caritas Est and Pope Francis’s writings on Christian love, and will host the Catholic AI Summit on March 19 to 20, 2027 at the Duquesne Club in Pittsburgh featuring D.C. Schindler, Charles Camosy, and L.M. Sacasas.
Why effective altruism is bad: the verdict from outside
Effective altruism draws criticism because its strict utilitarian spreadsheet logic turns human ethics into an elite optimization game that breaks down across five documented failure modes. While the movement’s premise of applying evidence and cost-benefit reasoning to avoid wasteful charity sounds reasonable on paper, stress-testing institutional EA against its own claims exposes where the architecture fails:
- The demandingness gap: Singer’s 1972 pond logic makes even a cinema ticket morally suspect and opens his 2015 book with kidney donation to strangers, forcing an institutional retreat to Giving What We Can’s negotiated 10% lifetime income pledge.
- Quantification bias: Reducing incommensurable human goods to QALYs and restaurant shrimp-suffering offsets filters out unmeasurable structural economic or tax reform, exemplified by the late-2022 Kurzgesagt controversy over Gates Foundation-sponsored “vote with your wallet” centrism.
- Longtermist priority drift: Silicon Valley tech wealth helped shift institutional EA from the global poverty focus of MacAskill’s Doing Good Better to the trillions of nonexistent future people in What We Owe the Future. Institutional EA now spends millions on speculative AGI extinction risks and feeds an Oxford talent pipeline straight into OpenAI and Anthropic.
- Top-down donor power and local neglect: Distant billionaires define the good without local accountability, whether in flops like TOMS Shoes and PlayPumps or in Kenya contraception campaigns. Institutional EA maintains a roughly 80/20 foreign-to-local ratio even as its San Francisco hub spends $80,000 to $90,000 per homeless person annually on streets where 100% of surveyed acquaintances wouldn’t walk alone at night.
- The self-defeating earn-to-give pipeline: Routing young idealists through profit-maximizing corporate careers erodes charitable dispositions, justifies present wealth like televangelism, and suffered its defining collapse with Sam Bankman-Fried and FTX in late 2022.
Looking at those five structural failures from outside the movement, I’m glad I’ve deliberately never donated to EA, attended its events, or joined its communities. The worldview feels like an unhealthy death cult because its culture trains people to weigh a Picasso or a farm full of shrimp against a drowning child, and to defer the real needs of living human beings for trillions of hypothetical future ones, not because anyone objects to doing math before donating.
You don’t need to join an emotionally unattuned utilitarian subculture or trust a crypto billionaire to care about whether aid actually works. Cheap malaria nets still deliver roughly a 100x impact multiplier over wasteful charity, GiveDirectly’s ~$50 monthly cash transfers still buy waterproof roofs for people in Africa living on $2 a day, and basic impact-optimizing math would be reinvented tomorrow if institutional EA disappeared, which is why the indictment of effective altruism lands not on the bed nets, but on what the movement’s spreadsheet ideology and billionaire funding turned it into.
Frequently Asked Questions
What are the downsides of altruism when it is measured by cost-effectiveness?
Treating ethics like a single continuous spreadsheet variable breaks down the moment you face human values that cannot be quantified. That quantification bias leads to absurd moral arithmetic, like calculating whether improving slaughter conditions for 2,000 shrimp per dollar outweighs saving a human child, while simultaneously filtering out systemic tax or economic reforms simply because their returns cannot be neatly modeled.
What are the criticisms of effective altruism?
The major failures come down to five structural bugs: an impossible moral standard that makes buying a movie ticket feel unethical, a quantification bias that ignores systemic reform, and a billionaire-backed pivot from poverty relief to speculative AI doom. On top of that, distant elites end up defining what counts as good without local accountability, and the movement’s earn-to-give pipeline routes idealistic kids into cutthroat corporate jobs that erode their actual generosity.
Is effective altruism a death cult?
It is not a literal religious cult, but its hyper-impartial utilitarian culture is the exact reason I have stayed completely out of the movement. When a philosophical framework trains people to lacks emotional attunement, argue for rescuing a Picasso over a child in a burning building, or prioritize trillions of nonexistent future people over living humans, it starts feeling deeply unhealthy from the outside.
Why did effective altruism turn to AI doomerism instead of helping people?
The shift happened when Oxford’s longtermist math met a massive influx of Silicon Valley tech money. By assuming that trillions of hypothetical future humans matter just as much as people alive today, figures like William MacAskill pivoted from the anti-poverty focus of Doing Good Better to the existential risk focus of What We Owe the Future, pouring millions into AGI alignment and feeding talent straight into labs like OpenAI and Anthropic.
Does effective altruism ignore local problems like San Francisco homelessness?
Institutional EA maintains roughly an 80% foreign to 20% local focus ratio, which leaves a jarring blind spot in the San Francisco Bay Area where EA has its highest per-capita membership. Even though the city reportedly spends $80,000 to $90,000 per unhoused person each year and surveyed locals say they will not walk the streets alone at night, defenders write off local misery as a governance failure where individual charity dollars cannot compete with overseas cost-effectiveness.
