Latest / I Believe / The Forethought of Grief

Transcript

The Forethought of Grief · I Believe

• 0:03
  SPEAKER_00: Welcome to I Believe, where we talk about America.
• 0:07
  SPEAKER_00: I'm Joel Douglas, and I believe in America.
• 0:10
  SPEAKER_00: Thanks for joining me.
• 0:32
  SPEAKER_00: Or maybe the press makes it seem like everybody is scared of artificial intelligence.
• 0:37
  SPEAKER_00: AI is going to take our jobs.
• 0:39
  SPEAKER_00: AI is going to replace artists.
• 0:42
  SPEAKER_00: AI is going to manipulate elections.
• 0:44
  SPEAKER_00: AI is going to become smarter than us.
• 0:47
  SPEAKER_00: Eventually, AI is going to destroy civilization.
• 0:50
  SPEAKER_00: Why are we always doing this?
• 0:53
  SPEAKER_00: Why do human beings always think civilization is about to end?
• 0:57
  SPEAKER_00: In my lifetime, first it was nuclear war.
• 1:00
  SPEAKER_00: We are all going to die.
• 1:02
  SPEAKER_00: Then the millennium change.
• 1:04
  SPEAKER_00: Everything is going to stop working.
• 1:06
  SPEAKER_00: Computers are going to fail.
• 1:07
  SPEAKER_00: Civilization is going to fall apart.
• 1:10
  SPEAKER_00: Then climate change.
• 1:11
  SPEAKER_00: Now AI is going to destroy us all.
• 1:14
  SPEAKER_00: Before that, people thought Haley's comet was going to poison everybody.
• 1:18
  SPEAKER_00: In 1910, scientists detected cyanogen gas in the comet's tail.
• 1:23
  SPEAKER_00: Earth was going to pass through it.
• 1:26
  SPEAKER_00: People bought gas masks and supposed remedies.
• 1:29
  SPEAKER_00: Newspapers ran with it.
• 1:30
  SPEAKER_00: There was real dread that this thing in the sky might kill everybody.
• 1:34
  SPEAKER_00: Then Earth passed through the tail, and nothing happened.
• 1:38
  SPEAKER_00: That does not mean every warning is stupid.
• 1:41
  SPEAKER_00: Nuclear war was a real danger.
• 1:43
  SPEAKER_00: It still is.
• 1:45
  SPEAKER_00: Y2K was a real computer problem.
• 1:47
  SPEAKER_00: People spent years fixing it.
• 1:50
  SPEAKER_00: Climate change is real.
• 1:51
  SPEAKER_00: AI is going to create real problems.
• 1:54
  SPEAKER_00: But we have this habit of moving really quickly from this could be bad to this could end civilization.
• 2:01
  SPEAKER_00: And I keep coming back to the same thing.
• 2:04
  SPEAKER_00: Every day has enough worry of its own.
• 2:06
  SPEAKER_00: Do not tax your life with the forethought of grief.
• 2:10
  SPEAKER_00: Enjoy the hard work of staying alive while you can.
• 2:13
  SPEAKER_00: There is that Wendell Berry poem, The Peace of Wild Things.
• 2:17
  SPEAKER_00: I am paraphrasing it, but it is something like: when I wake at night and worry about my life or my children's future, I go lie down by the water, among the wild things who do not tax their lives with the forethought of grief.
• 2:31
  SPEAKER_00: That phrase is just right.
• 2:33
  SPEAKER_00: The forethought of grief.
• 2:35
  SPEAKER_00: We do that all the time.
• 2:37
  SPEAKER_00: We grieve things that have not happened yet.
• 2:40
  SPEAKER_00: We suffer tomorrow before tomorrow gets here.
• 2:43
  SPEAKER_00: What happens to my children?
• 2:45
  SPEAKER_00: What happens to the climate?
• 2:47
  SPEAKER_00: What happens if the Russians launch?
• 2:49
  SPEAKER_00: What happens if the computers stop working?
• 2:52
  SPEAKER_00: What happens if the machine becomes smarter than us?
• 2:55
  SPEAKER_00: And some of those are reasonable questions.
• 2:58
  SPEAKER_00: But at some point, you are just paying today for grief that may never come.
• 3:02
  SPEAKER_00: That does not mean do nothing.
• 3:05
  SPEAKER_00: Take climate change.
• 3:06
  SPEAKER_00: Climate change is not going to kill every person on Earth.
• 3:10
  SPEAKER_00: Do you know what might kill a lot of people?
• 3:12
  SPEAKER_00: Doing nothing to prepare for climate change.
• 3:15
  SPEAKER_00: If we need to shore up buildings, improve infrastructure, strengthen the electrical grid, prepare for drought, fires, flooding, storms, then do those things.
• 3:25
  SPEAKER_00: That is useful.
• 3:26
  SPEAKER_00: Sitting around worrying that civilization is going to collapse is not useful.
• 3:31
  SPEAKER_00: Worry is not preparation.
• 3:33
  SPEAKER_00: And I think AI is the same way.
• 3:36
  SPEAKER_00: There are reasonable guardrails we can put around artificial intelligence without acting like the machine itself is evil.
• 3:43
  SPEAKER_00: Start with personal responsibility.
• 3:46
  SPEAKER_00: No fake lewd photographs or videos using another person's likeness.
• 3:50
  SPEAKER_00: That should be illegal, not because AI made the image, because you used somebody else's face to hurt them.
• 3:57
  SPEAKER_00: If an AI system hacks another company and causes millions of dollars in damage, somebody is responsible for that.
• 4:04
  SPEAKER_00: The AI did it cannot become some excuse that means nobody is liable.
• 4:09
  SPEAKER_00: A person deployed it.
• 4:10
  SPEAKER_00: A company built it.
• 4:12
  SPEAKER_00: Somebody authorized it to do something.
• 4:15
  SPEAKER_00: Now, I do not think the AI company should automatically be responsible every time somebody misuses a model.
• 4:21
  SPEAKER_00: That would be ridiculous.
• 4:23
  SPEAKER_00: If the company demonstrated reasonable guardrails, reasonable security, reasonable efforts to prevent abuse, then no, they should not automatically be liable because somebody found a way around those protections.
• 4:36
  SPEAKER_00: But if they knew there was a problem, knew their model was being used to do something harmful, had reasonable ways to stop it, and ignored it, then responsibility starts to move back toward the company.
• 4:47
  SPEAKER_00: That would make AI companies pay attention really quickly.
• 4:51
  SPEAKER_00: Not because somebody wrote a vague ethics statement, but because there would be consequences, there would be damages, there would be liability.
• 4:59
  SPEAKER_00: And in a messed up twist, I am not against the use of AI for military action.
• 5:04
  SPEAKER_00: I do not think that is contradictory.
• 5:07
  SPEAKER_00: The military already uses force that would be illegal in ordinary civilian life.
• 5:12
  SPEAKER_00: The question is, who is responsible?
• 5:14
  SPEAKER_00: If an autonomous or semi-autonomous system strikes a target, somebody in the chain of command still owns that decision.
• 5:22
  SPEAKER_00: The algorithm decided cannot be an excuse.
• 5:25
  SPEAKER_00: Somebody authorized that system.
• 5:28
  SPEAKER_00: Somebody gave it the authority to act.
• 5:30
  SPEAKER_00: Somebody remains responsible.
• 5:33
  SPEAKER_00: That gets into the philosophy behind all of this.
• 5:40
  SPEAKER_00: You are free to choose, so you are also responsible for what you choose.
• 5:44
  SPEAKER_00: You do not get to hide behind the system, or the institution, or the role, or eventually, the algorithm.
• 5:52
  SPEAKER_00: John Stuart Mill fits here too.
• 5:55
  SPEAKER_00: Mill's basic idea was that people should be left free to act unless their actions harm somebody else.
• 6:01
  SPEAKER_00: The state should not interfere just because an action is unusual, risky, unpopular, or offensive.
• 6:08
  SPEAKER_00: The line is harm.
• 6:10
  SPEAKER_00: That is pretty close to how I think about AI.
• 6:13
  SPEAKER_00: Do not regulate it because it frightens people.
• 6:15
  SPEAKER_00: Do not regulate it because it is new.
• 6:18
  SPEAKER_00: Do not regulate it because somebody can imagine a terrible future.
• 6:22
  SPEAKER_00: Regulate conduct when it causes real harm to other people.
• 6:26
  SPEAKER_00: That is a much cleaner rule.
• 6:28
  SPEAKER_00: If someone uses AI to commit fraud, punish the fraud.
• 6:33
  SPEAKER_00: If someone uses AI to create a nonconsensual sexual image, punish that.
• 6:38
  SPEAKER_00: If a company negligently deploys a system that causes foreseeable damage, hold the company responsible.
• 6:45
  SPEAKER_00: But do not build a giant apparatus around every hypothetical thing the technology might someday do.
• 6:51
  SPEAKER_00: You will never keep up anyway.
• 6:53
  SPEAKER_00: And I think the same principle applies to corporations.
• 6:56
  SPEAKER_00: Corporations have rights.
• 6:58
  SPEAKER_00: They can own property, they can enter contracts, they can sue and be sued.
• 7:04
  SPEAKER_00: They have legal protections.
• 7:05
  SPEAKER_00: Fine.
• 7:06
  SPEAKER_00: Then they also have reasonable responsibility.
• 7:09
  SPEAKER_00: You do not get rights without responsibility.
• 7:13
  SPEAKER_00: That applies to individuals.
• 7:14
  SPEAKER_00: It should apply to companies.
• 7:16
  SPEAKER_00: So maybe the philosophy is something like this maximum freedom with responsibility for foreseeable harm.
• 7:24
  SPEAKER_00: Let people build things.
• 7:25
  SPEAKER_00: Let companies innovate.
• 7:27
  SPEAKER_00: Let people use powerful tools.
• 7:30
  SPEAKER_00: But when somebody gets hurt, ask the same questions we already know how to ask.
• 7:35
  SPEAKER_00: Who did it?
• 7:36
  SPEAKER_00: Who authorized it?
• 7:38
  SPEAKER_00: Who could reasonably foresee the harm?
• 7:40
  SPEAKER_00: Who failed to take reasonable precautions?
• 7:43
  SPEAKER_00: Fraud is still fraud.
• 7:45
  SPEAKER_00: Theft is still theft.
• 7:47
  SPEAKER_00: Defamation is still defamation.
• 7:50
  SPEAKER_00: Negligence is still negligence.
• 7:52
  SPEAKER_00: Calling it artificial intelligence does not make responsibility disappear.
• 7:57
  SPEAKER_00: And that brings me back to the larger point.
• 7:59
  SPEAKER_00: Maybe AI could someday create some civilization level danger.
• 8:04
  SPEAKER_00: I do not know.
• 8:05
  SPEAKER_00: But every generation seems convinced it is standing at the edge of the end.
• 8:10
  SPEAKER_00: Nuclear war, the millennium, climate catastrophe, artificial intelligence.
• 8:15
  SPEAKER_00: Before that, Halley's comet.
• 8:18
  SPEAKER_00: At some point, you have to live.
• 8:20
  SPEAKER_00: If there is something useful to do, do it.
• 8:23
  SPEAKER_00: If the climate is changing and infrastructure needs to be stronger, build stronger infrastructure.
• 8:29
  SPEAKER_00: If AI can be used to make fake sexual images of people, make that illegal.
• 8:35
  SPEAKER_00: If AI systems can cause real damage, establish liability.
• 8:39
  SPEAKER_00: If companies put reasonable guardrails in place, do not punish them automatically because somebody misused the tool anyway.
• 8:47
  SPEAKER_00: Do the work that needs doing.
• 8:49
  SPEAKER_00: Fix what can be fixed.
• 8:51
  SPEAKER_00: Prepare where preparation actually matters.
• 8:54
  SPEAKER_00: And then live.
• 8:56
  SPEAKER_00: Because every day already has enough trouble of its own.
• 8:59
  SPEAKER_00: There are people to love, there are things to build, there are things to repair, there are books to read, there is dinner to make, there is work to do.
• 9:09
  SPEAKER_00: There is a world that is still here this morning.
• 9:12
  SPEAKER_00: Do not tax your life with the forethought of grief.
• 9:15
  SPEAKER_00: Enjoy the hard work of staying alive while you can.
• 9:19
  SPEAKER_00: Tomorrow has enough worry of its own.