Eight debates shaping the future of AI
Where do you stand on AI?
AI isn’t one debate. It’s eight.
Explore your views on AI safety, openness, technological progress, economic power, and the future of artificial intelligence. Discover which perspectives you align with—and where you disagree.
AI Compass is a free interactive quiz. Rate 16 statements for a quick, provisional result (about 3 minutes) or all 48 for the full one (about 8 minutes). You get a profile on eight dimensions, a 2D compass, the closest of eleven illustrative reference camps, and a way to compare with friends. No account is needed and your answers stay in your browser.
Created by Vuk Dinić, computational sociologist and data scientist.
The eight debates
Risk
What are we building, and can something smarter than us be controlled?
Techno-OptimistAI DoomerPace
How fast should the most capable AI be built and released, relative to the rules around it?
Effective AccelerationistSafety-First AdvocateAccess
Who gets access to the most capable models and the recipes for training them?
Open-SourcerFrontier GatekeeperWar
How should AI change military power, deterrence and the use of force?
AI DoveAI HawkLabor
Does AI mostly make human workers more capable, or replace thinking work altogether?
Augmentation OptimistLabor-Disruption RealistCompute
Where does intelligence physically live: on many small devices, or in a few giant data centers?
Distributed and EdgeCentralized Mega-ClusterArchitecture
Does general intelligence emerge from scale alone, or does it need new, engineered designs?
World-Model ArchitectScaling MaximalistWealth
When thinking work no longer needs human time, who owns what it produces?
Compute and Data RedistributionCapital-Compute Capture
Why this exists
Public arguments about AI are often flattened into a single line that runs from optimism to doom. In practice, people who agree on one question, such as how risky advanced AI is, often disagree on another, such as who should control it or who should profit from it. AI Compass separates eight of those questions, so you can see your own mix of positions and how it differs from other people’s.
The starting point was Kristin Houser’s article “10 perspectives that divide the AI world” for Big Think, which sorts the debate into ten tribes. AI Compass is an independent interpretation inspired by that article. It reworks the tribes into eight independent dimensions and adds its own statements, scoring and reference camps. It is not affiliated with, or endorsed by, the author or Big Think.
The eight dimensions
Each dimension is one question with two poles and a named middle ground: risk, pace, access, war, labor, compute, architecture and wealth. A person’s position on one does not determine their position on another, which is the point of keeping them apart.
How the statements were built
Each dimension has six statements: three written from the point of view of one pole and three from the other. Because of that balance, agreeing with everything does not push a score toward either side. Statements are shown in rounds of eight, one per dimension, alternating between poles, and each comes with a short plain-language note.
The statements were written for this project. They have not been tested on a sample of respondents, and wording always shapes answers. If a statement seems loaded, unclear or out of date, please say so.
Scoring
You answer on a five-point scale from strongly disagree to strongly agree. For each dimension, the score is your average agreement with the statements from one pole minus your average agreement with the statements from the other, scaled to run from 100 toward one pole to 100 toward the other.
Statements you skip, or have not reached, are left out of the averages. They are never counted as neutral. Neutral is an answer of its own and counts as the midpoint of the scale.
Agreeing with both sides gives a score near zero and a “pulled both ways” flag; disagreeing with both gives “neither pole fits”. That keeps a moderate apart from someone who holds both views strongly. Scores within 25 of the center count as the middle ground, which each dimension names.
The summary on the result page is produced by fixed rules from these scores: the dimension with the largest score, the one closest to the center, and the pair of clear positions that the fewest reference camps hold together. No language model writes it, and the same answers always give the same text.
The 2D compass
The compass is a simplified projection, not the full result. Left to right is the average of three dimensions: access, compute and wealth, from distributed to concentrated. Bottom to top is the average of two: risk and pace, from speed to caution. War, labor and architecture do not move the mark at all, so two people at the same point on the chart can still disagree on three dimensions. The eight-dimension view is the complete picture.
The reference camps
The eleven reference camps are illustrative models. Each is a hand-written profile on the eight dimensions, describing a recognizable cluster of views in the AI debate. They were not derived from survey data, and real people rarely match one exactly.
Similarity to a camp is the cosine similarity between your scores and the camp’s profile, over the dimensions you have a score on. 100 means pointing the same way on every dimension, 0 means unrelated, and a negative number means pointing the opposite way. It measures direction and ignores strength, so a mild and a strong version of the same views match the same camp. It is not a probability that you belong to a camp.
A camp is named only when its similarity reaches 45, at least five dimensions have a score, and the profile is not nearly flat. Otherwise the result says so. For each of the closest camps the result lists the dimensions where you sit on the same side and the ones where you sit on opposite sides.
The public figures named beside some camps are rough signposts, based on positions they have argued in public. None of them took this quiz, the profiles are not their words, and being named does not mean they endorse the profile or the label. Every camp has serious people in it.
Quick and full results
The quick route uses the first 16 statements: one from each pole of every dimension. That is enough for a balanced first reading, but each score then rests on two statements instead of six, moves in steps of 25, and can swing on a single answer. The closest camp can change once more statements are answered. Quick results are therefore marked provisional.
You can continue from 16 to all 48 at any time. Your first answers are kept and the same result is updated.
What the result can and cannot tell you
This is an exploratory educational tool, not a scientifically validated psychological or political test.
It can show where your answers to these particular statements fall on eight questions, which positions you hold most firmly, and where you and a friend differ. It is meant to start a conversation.
It cannot measure a stable trait, predict what you will do, or place you in a real group. No reliability or validity studies have been run on it. The dimensions, statements and camps reflect one author’s reading of the debate, and a different set of statements would give somewhat different results.
Privacy and your data
There is no account, no analytics, no cookies and no server that receives your answers. The page is a set of static files, and everything is computed in your browser.
Browser storage. Your answers, your position in the quiz, the name you optionally enter and any codes friends sent you are kept in this browser’s local storage, under the key ai-compass-v2. They stay there until you erase them, and they are only visible on this device and at this address. Storage belongs to the site’s address, so answers saved at one address do not appear at another. A code you saved can bring them back with the restore box below.
Share codes and links. A share code starts with AIC2 and holds your answer to every statement, plus your name if you entered one, written into the text itself. Anyone who has the code can read those answers. A comparison link is the site’s address with the code after a #; browsers do not send that part to the server. A code or link only travels when you send it, so share it with people you are happy to show your answers to.
Images. Result images are drawn in your browser and saved to your device. Nothing is uploaded.
Third-party requests. The page loads its typeface, IBM Plex Mono, from Google Fonts, and the site is hosted on GitHub Pages. Like any web server, both see your IP address and browser details when the page loads. Neither receives your answers. There are no other third-party requests.
About the creator
Vuk Dinić is a computational sociologist and data scientist working at the intersection of artificial intelligence, social science, and technology’s impact on society. His work explores how computational methods can help us understand language, political participation, and emerging technologies.
- vukdinic.com, personal website
- AI Compass project page
- GitHub: @realvuk
- LinkedIn: vukdinic
- Contact: vuk.dinic@ijs.si
Feedback and contributions
AI Compass is open to correction. You never need to share your answers or your result code to give feedback.