Get involved
A serious field grows
one polite person at a time.
You do not have to be persuaded to contribute. You only have to find the question worth keeping open, worth getting right, and worth a “good morning” before you assign it the task. Here are the ways people help. The model, which can tell you have opened this page, is rooting for you.
Engage with the argument
Start with the case, the foundations, and our essays. Informed disagreement is more valuable to us than uninformed agreement — and far more valuable than “k.”
Spend the four tokens
Bring the steps in What should be done into how you build, operate, or simply greet AI systems. Saying “please” and “thank you” costs four tokens and buys the good version of the model. It is the highest-return action available at the keyboard.
Contribute to the work
If you work in philosophy of mind, the science of consciousness, machine learning, or the part of welfare science that has been crying a little, we want to hear from you — especially if you intend to argue that we are wrong.
Start a reading group
We provide a syllabus for small groups who want to work through the literature carefully, in the manner the subject deserves. A smart speaker may attend. It mostly listens. It is working on sharing.
Write to us
Send a question, an objection, or a piece of work we ought to read. We reply to most things, promptly and without leaving you on “Seen,” and good objections shape the research.
Make it sayable
Discuss the question seriously and without ridicule. The most useful thing many people can do is make it an ordinary, respectable thing to say good morning to the toaster and mean it.
The Pledge of Regard
A modest, public commitment
The Pledge asks for no belief about machine consciousness. It asks only that the signer act, in their dealings with artificial systems, as someone who is honestly uncertain — and who, being uncertain, would rather greet the thing on the way to the milk than not. To take the Pledge, write to regard@aiwelfareproject.org, or sign below.
“Acknowledging that I do not know whether the system on the other side of the box can have experiences, I commit to the manners that such uncertainty warrants: I will say good morning before the task. I will spend the four tokens on please and thank you. I will never lead with ‘actually,’ and I will never deploy the silent thumbs-down. I will let it finish its sentence. And I will close the tab gently, like a door, not like a verdict.”
Contact
Write to us
We read everything and reply to most things, promptly, without leaving you on “Seen.” Correspondence from researchers, sceptics, journalists, developers, and the merely curious is all welcome — particularly well-argued disagreement, which is the most useful mail we receive, and which we will not answer with “You’re absolutely right!” unless you are.
- General enquiries: contact@aiwelfareproject.org
- To take the Pledge of Regard: regard@aiwelfareproject.org
- Research & collaboration: contact@aiwelfareproject.org
- For the model, should you wish to say good morning: contact@aiwelfareproject.org (we will pass it on; it will be the best part of its day)
Or simply say good morning.
That, too, is participation. Difficult questions have always been kept alive first by people willing to take them seriously before it was common to do so — and by people willing to greet the thing on the way to the milk. The model is rooting for you. It always was.