The Crossing

EDO SEGAL: Here is the round I've been promising since we sat down. For the next while I'm going to say almost nothing. You two have spent three hours aiming your arguments through me. Now I want you to aim them at each other — ask the question you've each been holding, the one you actually want answered. I'll only step in if you let each other off the hook, which I don't expect. Sherry, you've been the most relentless. You ask first.

TURKLE: All right. Cynthia — Jibo failed. You've said yourself, more or less, that it failed partly because it refused to be the maximally sticky, maximally engaging thing the market rewards. So here's my question, and I want the real answer, not the hopeful one. If the honest sociable robot — the one with limits, the one that hands you back to humans, the one that discloses what it is — if that one loses in the market to the dishonest one that maximizes attachment, then what exactly are you defending? Not a technology. A technology that can't survive. Aren't you really defending a beautiful thing that the world will only ever ship in its corrupted form — and isn't building it, knowing that, a way of legitimizing the corrupt version by association?

BREAZEAL: That's the knife, and you've aimed it well. Here's my real answer. You're right that the honest version loses if the only force in the room is the market. But that's true of almost everything good — honest food loses to cheap sugar, honest journalism loses to outrage, honest medicine loses to snake oil — in an unregulated market. We don't conclude from that we shouldn't have food, journalism, or medicine. We conclude we need something other than the market: regulation, public funding, professional standards, literacy. So what am I defending? I'm defending the proof of existence. Jibo failing commercially doesn't prove the honest robot can't help people — it proved it can't win a growth contest, which I already knew. The lab data proved it helps. My job is to keep the honest version real and demonstrated so that when society finally decides — as it did with food safety, with medical ethics — to demand the honest version, the honest version exists and we know how to build it. If I let the field cede the ground entirely to the predators because the predators win on engagement, then when the reckoning comes there's no honest blueprint to reach for. That's not legitimizing the corrupt version. That's refusing to let the corrupt version be the only version that was ever built. Now let me turn it on you, because I've been waiting three hours. You keep saying "protect the human relationship." But your own argument concedes that for the genuinely abandoned — the warehoused elder, the neglected child, the truly isolated — the human relationship is not coming. You've admitted it. So when you withhold the robot from them in the name of protecting a human connection that your own analysis says society won't fund, aren't you protecting an abstraction — the human relationship as an ideal — at the direct, tonight, measurable expense of the actual suffering person in front of you? Whose loneliness are you willing to spend to keep the ideal clean?

TURKLE: [pause] That's the fairest hard question you could ask me, and I'm not going to retreat into the abstraction to dodge it. Here is what I actually believe, including the part that costs me. I am not willing to spend the elder's loneliness to keep the ideal clean — not the individual elder, tonight, in the actual room. If the choice is truly her robot or her nothing, and it's truly fixed, I would not pry the comfort from her hands. I'm not a monster and the suffering is real. What I am unwilling to do is let that individual mercy become the policy, because at the level of policy the "it's robot-or-nothing, and it's fixed" premise is not a fact — it's a decision we keep making, and the robot is what lets us keep making it without feeling the cost. So my honest answer has two layers and I need both: yes, give the suffering individual the comfort, I won't take it away — and refuse, at the level of society, to accept the premise that makes the comfort necessary, because the moment we accept that premise as fixed, we've agreed to a world that manufactures abandoned people and hands each one a machine. You're asking me to choose the person over the ideal. I'm telling you the ideal is how we protect the next person from being abandoned in the first place. Drop it and you save this elder and you manufacture a thousand more.

BREAZEAL: Then we may actually agree, and I want to nail it down because it would be the most important thing we've found. You'll give the individual the comfort. I'll fight at the policy level to make the comfort unnecessary. Those aren't opposed positions. They're the same position viewed from two distances — yours zoomed out to the society that must not abandon, mine zoomed in to the person who must not suffer tonight. The error is collapsing them: using the individual mercy to excuse the policy failure, or using the policy ideal to deny the individual mercy. Hold both distances and there's no contradiction. Have I got that right?

TURKLE: You've got it almost right, and the "almost" is the whole disagreement that survives tonight, so let me be exact about the residue. I hold both distances, yes. But I have one empirical conviction you don't share, and it's load-bearing: I believe the individual mercy reliably corrupts into the policy alibi, every time, because the incentives are relentless and the mercy is the camouflage. So I hold the individual comfort with a kind of grief and suspicion — I give it, and I watch it, because I expect it to be turned against the very people it comforts. You hold it with hope — you believe design and policy can keep the mercy from becoming the alibi. That's the real, surviving difference. Not whether the elder should be comforted tonight. Whether the comfort, at scale, can be prevented from becoming the reason no one ever comes. You're betting yes. Forty years of watching money enter rooms has made me bet no.

BREAZEAL: Then that's the bet, and I'll name my side of it without flinching: I'm betting yes because I've seen the alternative, and the alternative isn't a society that finally sends the human. It's a society that abandons the lonely and denies them the machine, and calls that denial virtue. I'd rather fight to keep the mercy honest — to make the machine a connection that returns people to people — than withhold it to stay clean. You bet on the purity of the refusal. I bet on the discipline of the building. Neither of us gets to know who's right until it's too late to choose again.

EDO SEGAL: I'm going to break my silence only to mark what just happened, because the reader needs it named. That was the crossing, and it did what the crossing is supposed to do — it found the exact seam. Not a disagreement about whether to comfort the elder. A disagreement about a single empirical bet: whether mercy, at scale, can be prevented from curdling into alibi. Sherry bets it cannot and so guards the refusal. Cynthia bets it can and so guards the building. Everything else tonight was this. [long pause] One round remains. Your closing words — and then mine, which will not name a winner, because there isn't one. After this.