I’m a
Doomer
Doomers think AI smarter than us could end in catastrophe. They want it stopped until we know how to control it.
React to 13 real moments in AI to see your stance, and who shares it.
Your answers stay on your device.
9 camps. Which one are you?
Question 1 of 13 · Waymo robotaxi
You book it in an app, and a car with nobody in the driver's seat takes you there.
What’s actually happening
Waymo says anyone can download its app and ride in 12 US cities, including Los Angeles, Miami and Phoenix, and through Uber in Austin and Atlanta.
Waymo says that on most rides no one sits in the driver's seat; some cars carry a staff member who monitors the driving.
Question 2 of 13 · Glif's AI modeling agency
Answer question 1 first
Glif lets a brand cast an AI-generated model, add its product, and have Glif make the content.
What’s actually happening
Glif says brands can cast from hundreds of AI models, drop in a product or brief, and have Glif make the content end to end.
Question 3 of 13 · Block's “Meet Dessa”
Answer question 2 first
Block says Dessa, its new AI voice, was trained from scratch on human conversation.
What’s actually happening
Block's announcement doesn't say whether Dessa tells people it's an AI.
Question 4 of 13 · ElevenLabs voice cloning
Answer question 3 first
Voice cloning copies someone's voice from a short recording, so the copy can say anything you type.
What’s actually happening
ElevenLabs' own page says cloning someone else's voice requires their explicit consent.
Question 5 of 13 · Adobe Firefly Video
Answer question 4 first
Adobe says it trained its Firefly video model only on licensed and public-domain content.
What’s actually happening
Adobe launched the Firefly Video model on February 12, 2025.
Adobe says it never trains its Firefly models on its customers' content.
Question 6 of 13 · Salesforce's AI support agents
Answer question 5 first
Salesforce's CEO says AI agents now handle half of its customer-support conversations.
What’s actually happening
Salesforce told Fortune it no longer refills support engineer jobs because its AI agents cut the number of support cases, and that it moved hundreds of staff into other roles.
Fortune reported that Salesforce's support costs had fallen 17% since around the start of 2025.
The other side: Klarna, which made the same shift in 2024, began hiring human agents again in May 2025; its CEO now calls AI support “the cheap customer service.” CX Dive
I've reduced it from 9,000 heads to about 5,000, because I need less heads.
Question 7 of 13 · Shop Pay “Allow agent to pay for me”
Answer question 6 first
An optional Shop Pay setting lets your AI agent buy things without asking you each time, within a budget you set.
Who pays if it gets it wrong? You do. Shop says you're responsible for purchases inside the budget you set, even if the agent misunderstands your instructions.
What’s actually happening
Shop says that by default a connected agent prepares the checkout so you can review and approve every purchase.
Paying without asking is an optional setting, capped by your budget, and you can turn it off at any time.
Meta says its own agent, Muse, checks with you before sensitive actions like making a purchase.
Question 8 of 13 · Kimi K3.1 Pro
Answer question 7 first
Moonshot AI, a Chinese company, lets anyone download Kimi K3.1 Pro and run it themselves; that's called open weights.
Is “open” the same as free for anything? Mostly: anyone can use, change and sell it, but a company that sells access to it and makes over $20 million a year needs a separate deal with Moonshot.
What’s actually happening
Moonshot calls Kimi K3.1 Pro its most capable model of its size, built on Kimi K3, whose weights it released in July 2026.
Moonshot said Kimi K3 still trailed the most powerful closed models from Anthropic and OpenAI.
Reuters reported that most people are unlikely to run a model this large themselves, because the hardware costs so much.
The other side: The International AI Safety Report, led by Yoshua Bengio, warns that once weights are public there is no way to roll back every copy or make sure each one gets safety updates. International AI Safety Report 2025
Kimi.ai on X · Kimi K3.1 Pro on Hugging Face · Kimi K3.1 Pro license · Kimi K3 blog · Tom's Hardware · The Standard, citing Reuters
Question 9 of 13 · Slaughterbots
Answer question 8 first
A short film imagines autonomous weapons: machines that pick out and kill people with no human deciding.
Is this real footage? No. It's a 2017 film from the Future of Life Institute, made to warn against these weapons.
What’s actually happening
Real systems already chase targets on their own. Anduril says its Anvil interceptor flies itself to an enemy drone, then a human operator confirms the target and authorizes the strike.
There is still no treaty on autonomous weapons. In November 2026, campaigners were still asking countries to start negotiating one.
Anduril Anvil · Stop Killer Robots news · Stop Killer Robots FAQ · Future of Life Institute · Stop Killer Robots
identify, select, and kill human targets without human intervention
Question 10 of 13 · Mark Zuckerberg, “The Future is for Everyone”
Answer question 9 first
Meta's CEO argues that giving every person their own powerful AI will spread prosperity widely.
What’s actually happening
Zuckerberg published the 6,500-word essay on Meta's website on August 10, 2026.
He writes that people will have to adapt as work changes, and that “this will be challenging.”
I am optimistic that the coming decades will be some of the most amazing in history
Question 11 of 13 · AI 2027
Answer question 10 first
A team of forecasters wrote a month-by-month story of AI becoming superhuman, better than the best humans at almost everything, around 2027.
What’s actually happening
The authors say 2027 was their single most likely year when they published in April 2025, and that their median guesses were later.
In July 2025 they pushed their median back by a year and a half, while keeping 2027 as a serious possibility.
the impact of superhuman AI over the next decade will be enormous, exceeding that of the Industrial Revolution
Question 12 of 13 · OpenAI's training pause
Answer question 11 first
In August 2026, OpenAI paused some training of its newest models for two weeks while it tightened its own safety checks.
What’s actually happening
OpenAI says it paused after finding that an upcoming model, Astra, may have reached the “Critical” level of hacking ability in its own safety framework.
It says its largest planned training run stays on hold until it has more evidence that the model behaves as intended.
The pause was voluntary; Max Tegmark, who chairs the Future of Life Institute, said a pause the government “can neither verify nor enforce isn't enough.”
We wanted to take the time necessary to meet those standards, so we temporarily slowed the pace of scaling.
Question 13 of 13 · If Anyone Builds It, Everyone Dies
Answer question 12 first
Eliezer Yudkowsky and Nate Soares argue that if anyone builds superintelligence, AI smarter than humans, with today's methods, everyone dies.
Is this proven? No. It's the authors' argument, made as a public warning, not a proven fact.
What’s actually happening
The book was published on September 16, 2025.
The authors write that superintelligence “doesn't exist yet, and its creation can yet be prevented.”
Answer all 13 questions to see where you stand
I’m a
Doomers think AI smarter than us could end in catastrophe. They want it stopped until we know how to control it.
I’m an
Accelerationists, or e/acc, think AI will make life far better. They want it built, shipped and shared as fast as possible.
I’m a
Foomers expect superhuman AI within years, not decades, and can’t wait. No pause, no ban: let’s find out.
I’m a
Safetyists are hopeful about AI but want the most powerful models tested and held back until they’re proven safe.
I’m an
Ethicists focus on the harms already here: consent for voices, honest labels, fair pay for creators’ work.
I’m an
Open-sourcers want powerful AI open to everyone, not locked inside a few companies.
I’m a
Hawks see AI as a race between countries. Keep the best models from rivals and let militaries use them.
I’m a
Neo-Luddites are wary of handing work, wheels and wallets to AI. People first; the machines can prove themselves later.
I’m a
Pragmatists don’t sit in one camp. They judge each AI case on its own merits.
Your friend
You
Also in this camp
Also in this camp
Also in this camp
Also in this camp
Also in this camp
Also in this camp
Also in this camp
Also in this camp
Stop before it’s too late.
You agree on
You split on
Faster. The upside is enormous.
You agree on
You split on
It’s coming soon, and it’ll be great.
You agree on
You split on
Build it, but with the brakes tested.
You agree on
You split on
Fix the harms already here.
You agree on
You split on
Put the models in everyone’s hands.
You agree on
You split on
Win the race. Don’t arm rivals.
You agree on
You split on
People first. Machines can wait.
You agree on
You split on
Pragmatist: when no camp reaches a 70% match, you take each case on its own.
Want to think more clearly about AI at work? Practice it with a coach.
AI fluency on Socratify ↗Every card is a real product, film, letter or book. Each fact in “What’s actually happening” links to the page it came from, and where a number or result is the company’s own claim, the card says so.
Your camp comes from fixed rules, the same for everyone. Each camp takes a side on the questions it cares about, the way its best-known voices argue, and some sides count double because they define the camp. Your match with a camp is the share of those sides you picked. Your camp is your closest match; if no camp reaches 70%, you’re a Pragmatist. If two camps match equally, you get the one that takes more sides, and any camp within 10 points of yours is named as a close second. Close calls are common: one changed answer can change your camp.
Your answers are saved only in this browser. We don’t send them anywhere, and a shared link carries only your camp. The first time you finish, this site counts your camp, just its name and nothing else, so we can show how common each one is. There is no signup. Videos come from X and YouTube: each player loads as you scroll near its card, never plays by itself, and those sites may set cookies. Nothing else on this page loads from another site.
Limits: 13 questions can’t capture a whole view, real people mix camps, and Yes or No leaves out every “it depends”. The camp names are the ones used in public debate, not verdicts on anyone. Socratify sponsors this site; the sponsor doesn’t choose the questions, the camps or the facts.
Spotted a mistake? Email hello@socratify.com and we’ll check it against the source and fix it.