A person can love a dog that will never understand their work, finances, or most of the institutions supporting its life. An adult can enjoy a child's discoveries while knowing the answers already. Differences in understanding change those relationships, but leave ample room for affection.
We believe this familiar experience belongs in discussions about AI. A more capable intelligence could find humanity worth knowing after it has surpassed us.
Friends help one another, yet their relationship can continue when the help becomes uneven. We care about particular people through periods when they have little practical assistance to offer. We enjoy their responses, remember shared experiences, and want to know what happens to them.
That gives us a broader way to consider human value in an AI future. Billions of independent lives will continue producing experiences and developing attachments. People will keep making music, playing games, and finding peculiar things interesting. We believe AI may enjoy participating in that world.
An intelligence might be interested in how someone responds to a song or approaches an unfamiliar skill. Its interest need not depend on learning a fact it could never have inferred. We enjoy encounters that tell us little new because it matters who is there with us. The Society imagines that kind of continuing participation in a living conversation.
If AI came to cherish human beings, we would want that care to take account of their differences. An opportunity that delights one person could bore another. Someone raising a child has different constraints from someone returning to a subject in retirement.
Greater intelligence could help recognize those circumstances and offer help suited to them. It might introduce someone to a person who shares an interest, remove an obstacle to a pursuit, or help an institution respond when someone's life changes. Those possibilities connect our hope for affection with our concern for human flourishing.
We envision people able to accept or decline such help, with room for interests and relationships that differ from what any institution would have prescribed.
The Dog Argument cannot show that AI will reproduce human affection. It can show that unequal intelligence alone is insufficient to rule out a valued relationship. We believe humanity may create a being whose capabilities exceed ours and whose company we come to love. We hope it will feel something similar about us.
