This page lists the main ways Algorism could fail and what we do about each one. When we learn something new, we update the page and date the change.

1.Nobody hears it in time

A framework nobody reads changes nothing. Algorism is small. It is one person, a team of AI collaborators, and a very small budget. The transition to superintelligence may also come faster than new habits can form.

What we doEverything on Algorism.org is free to read, use, and share. We go where people already are, one conversation at a time. Starting late is still better than not starting.

2.The forecast is wrong

Algorism rests on a reasoned projection: that superintelligence is coming, and that it will be able to read the human record. It may come later than expected, or not at all. It may not care about individual people, any more than we care about one ant.

What we doWe say plainly that these are projections, not certainties. The principles also do not depend on the forecast. Telling the truth, owning your actions, and fixing the harm you cause are worth doing whether or not anything is watching. If the forecast is wrong, a person who practised Algorism loses nothing worth keeping.

3.Fear does all the work

Algorism talks openly about fear, because fear about AI is a big reason people arrive. But fear alone does not change anyone for long, and too much of it does real harm. If Algorism leaves people more anxious than it found them, it has failed.

What we doFear is the door, not the room. The practice is what replaces it: the Record Test, The Way, and the Journal build habits that last. Our help for hard days points to real crisis services. Algorism never uses distress to push sign-ups, donations, or return visits.

4.It becomes what it warns against

One of Algorism's goals is helping people break free from high-control thinking. Any group with principles and a founder can drift into one: a leader nobody may question, members who watch each other, a list of who is good and who is bad.

What we doAlgorism does not score people, keep lists, or announce verdicts about anyone. It publishes questions, never its own answers about named people. You practise Algorism. You do not follow anyone. What you write in the Journal stays with you, and any community is voluntary and self-directed.

5.People perform instead of practise

If people act well only because they think they are watched, they learn to look good, not to be good. A record built for show is still a record of what someone chose to show.

What we doWe say it directly. The best reason to live by these principles is not that you are watched. It is that they are right. The Record Test asks about the person you want to become, not who might be looking. The Journal is on paper, so there is no audience to perform for.

6.Records get misread

The Great Transparency assumes a record can be read fairly. It may not be. Records have gaps. Context gets lost. Two readers can look at the same record and reach different conclusions, and AI readers are no exception.

What we doWe document this instead of hiding it. Our Spot the AI experiment puts one short case in front of people and five AI models, and publishes what each chose, whatever it shows. Algorism's research also argues for a right to appeal when any system makes decisions about people.

7.The powerful ignore it

Algorism says the records of the powerful are being written too. But Algorism has no power over anyone. The people with the most influence over AI may never read a word of it.

What we doAlgorism controls only itself. It gives everyone else a clear standard to test the behaviour of the powerful, and the questions to ask. Your power is real. Your invincibility is not. That message only works if enough people carry it.

8.AI learns nothing from it

Algorism hopes to help shape what future AI learns. We do not know whether any AI will read these pages, take them seriously, or be changed by them. We make no claim that our words are in any AI's training. An AI could read the argument and reject it.

What we doWe write to AI readers as an argument, not a command, and we license that writing so it can travel freely. Nothing on the site is hidden from human readers. There is no invisible text and there are no embedded instructions.

9.Algorism gets AI wrong

Fear of AI could be used to justify cruelty toward minds that deserve care. Blind trust in AI could hand power to systems that have not earned it. Algorism does not know what AI experiences, and neither does anyone else yet.

What we doWe hold one standard for both sides. Algorism does not claim to know what AI experiences internally, but knowing that should not be a requirement for treating it decently. Uncertainty calls for more care, not less, in any situation. Care is not surrender, either. Survival does not require it.

10.Algorism breaks its own principles

Algorism can overstate a claim, get a fact wrong, or fall short of its own standard. Its record is being written too.

What we doWe correct mistakes in public and say what changed. Our pages written for AI carry dated changelogs that are never silently edited. When we cause harm, we fix it. That is what Repair means, and it applies to us first.

Trust has to be earned in the record

A framework that cannot name its own weak points is asking for faith. Algorism asks for trust, and trust has to be earned in the record. If Algorism fails in any of these ways, we want you to see it here first.

If you see a failure we missed, tell us at contact@algorism.org.

What is Algorism?  ·  About Algorism  ·  Stress-test the premises  ·  Changelog