[Script Info]
Title: 
[Events]
Format: Layer, Start, End, Style, Name, MarginL, MarginR, MarginV, Effect, Text
Dialogue: 0,0:00:01.29,0:00:03.17,Default,,0000,0000,0000,,It’s 6:30 in the morning,
Dialogue: 0,0:00:03.58,0:00:08.46,Default,,0000,0000,0000,,and Kristen is wheeling\Nher prostate patient into the OR.
Dialogue: 0,0:00:09.50,0:00:11.75,Default,,0000,0000,0000,,She's a resident, a surgeon in training.
Dialogue: 0,0:00:12.33,0:00:14.50,Default,,0000,0000,0000,,It’s her job to learn.
Dialogue: 0,0:00:15.29,0:00:18.64,Default,,0000,0000,0000,,Today, she’s really hoping to do\Nsome of the nerve-sparing,
Dialogue: 0,0:00:18.67,0:00:22.54,Default,,0000,0000,0000,,extremely delicate dissection\Nthat can preserve erectile function.
Dialogue: 0,0:00:23.50,0:00:26.84,Default,,0000,0000,0000,,That'll be up to the attending surgeon,\Nthough, but he's not there yet.
Dialogue: 0,0:00:27.62,0:00:30.02,Default,,0000,0000,0000,,She and the team put the patient under,
Dialogue: 0,0:00:30.04,0:00:33.75,Default,,0000,0000,0000,,and she leads the initial eight-inch\Nincision in the lower abdomen.
Dialogue: 0,0:00:35.04,0:00:38.63,Default,,0000,0000,0000,,Once she’s got that clamped back,\Nshe tells the nurse to call the attending.
Dialogue: 0,0:00:39.58,0:00:41.88,Default,,0000,0000,0000,,He arrives, gowns up,
Dialogue: 0,0:00:42.46,0:00:48.25,Default,,0000,0000,0000,,And from there on in, their four hands\Nare mostly in that patient --
Dialogue: 0,0:00:48.71,0:00:51.62,Default,,0000,0000,0000,,with him guiding\Nbut Kristin leading the way.
Dialogue: 0,0:00:52.88,0:00:57.52,Default,,0000,0000,0000,,When the prostates out (and, yes,\Nhe let Kristen do a little nerve sparing),
Dialogue: 0,0:00:57.54,0:00:58.77,Default,,0000,0000,0000,,he rips off his scrubs.
Dialogue: 0,0:00:58.79,0:01:00.17,Default,,0000,0000,0000,,He starts to do paperwork.
Dialogue: 0,0:01:00.83,0:01:06.21,Default,,0000,0000,0000,,Kristen closes the patient by 8:15,
Dialogue: 0,0:01:06.58,0:01:09.02,Default,,0000,0000,0000,,with a junior resident\Nlooking over her shoulder.
Dialogue: 0,0:01:09.04,0:01:12.12,Default,,0000,0000,0000,,And she lets him do\Nthe final line of sutures.
Dialogue: 0,0:01:12.83,0:01:15.88,Default,,0000,0000,0000,,Kristen feels great.
Dialogue: 0,0:01:16.25,0:01:17.81,Default,,0000,0000,0000,,Patient’s going to be fine,
Dialogue: 0,0:01:17.83,0:01:21.00,Default,,0000,0000,0000,,and no doubt she’s a better surgeon\Nthan she was at 6:30.
Dialogue: 0,0:01:22.21,0:01:25.04,Default,,0000,0000,0000,,Now this is extreme work.
Dialogue: 0,0:01:25.42,0:01:29.25,Default,,0000,0000,0000,,But Kristin’s learning to do her job\Nthe way that most of us do:
Dialogue: 0,0:01:29.62,0:01:31.52,Default,,0000,0000,0000,,watching an expert for a bit,
Dialogue: 0,0:01:31.54,0:01:34.68,Default,,0000,0000,0000,,getting involved in easy,\Nsafe parts of the work
Dialogue: 0,0:01:34.71,0:01:36.89,Default,,0000,0000,0000,,and progressing to riskier\Nand harder tasks
Dialogue: 0,0:01:36.92,0:01:39.25,Default,,0000,0000,0000,,as they guide and decide she’s ready.
Dialogue: 0,0:01:40.04,0:01:42.93,Default,,0000,0000,0000,,My whole life I’ve been fascinated\Nby this kind of learning.
Dialogue: 0,0:01:42.96,0:01:46.62,Default,,0000,0000,0000,,It feels elemental,\Npart of what makes us human.
Dialogue: 0,0:01:47.75,0:01:53.17,Default,,0000,0000,0000,,It has different names: apprenticeship,\Ncoaching, mentorship, on the job training.
Dialogue: 0,0:01:53.54,0:01:56.83,Default,,0000,0000,0000,,In surgery, it’s called\N“see one, do one, teach one.”
Dialogue: 0,0:01:57.62,0:01:58.97,Default,,0000,0000,0000,,But the process is the same,
Dialogue: 0,0:01:58.99,0:02:03.17,Default,,0000,0000,0000,,and it’s been the main path to skill\Naround the globe for thousands of years.
Dialogue: 0,0:02:04.33,0:02:08.83,Default,,0000,0000,0000,,Right now, we’re handling AI\Nin a way that blocks that path.
Dialogue: 0,0:02:09.62,0:02:12.32,Default,,0000,0000,0000,,We’re sacrificing learning\Nin our quest for productivity.
Dialogue: 0,0:02:13.29,0:02:16.10,Default,,0000,0000,0000,,I found this first in surgery\Nwhile I was at MIT,
Dialogue: 0,0:02:16.12,0:02:18.60,Default,,0000,0000,0000,,but now I’ve got evidence\Nit’s happening all over,
Dialogue: 0,0:02:18.62,0:02:22.50,Default,,0000,0000,0000,,in very different industries\Nand with very different kinds of AI.
Dialogue: 0,0:02:23.08,0:02:28.93,Default,,0000,0000,0000,,If we do nothing, millions of us\Nare going to hit a brick wall
Dialogue: 0,0:02:28.96,0:02:31.38,Default,,0000,0000,0000,,as we try to learn to deal with AI.
Dialogue: 0,0:02:33.12,0:02:34.90,Default,,0000,0000,0000,,Let’s go back to surgery to see how.
Dialogue: 0,0:02:35.71,0:02:37.64,Default,,0000,0000,0000,,Fast forward six months.
Dialogue: 0,0:02:37.67,0:02:43.14,Default,,0000,0000,0000,,It’s 6:30am again, and Kristen\Nis wheeling another prostate patient in,
Dialogue: 0,0:02:43.17,0:02:46.33,Default,,0000,0000,0000,,but this time to the robotic OR.
Dialogue: 0,0:02:47.67,0:02:49.35,Default,,0000,0000,0000,,The attending leads attaching
Dialogue: 0,0:02:49.38,0:02:52.21,Default,,0000,0000,0000,,a four-armed, thousand-pound\Nrobot to the patient.
Dialogue: 0,0:02:52.75,0:02:55.18,Default,,0000,0000,0000,,They both rip off their scrubs,
Dialogue: 0,0:02:55.21,0:02:58.33,Default,,0000,0000,0000,,head to control consoles\N10 or 15 feet away,
Dialogue: 0,0:02:59.17,0:03:02.92,Default,,0000,0000,0000,,and Kristen just watches.
Dialogue: 0,0:03:04.38,0:03:07.43,Default,,0000,0000,0000,,The robot allows the attending\Nto do the whole procedure himself,
Dialogue: 0,0:03:07.45,0:03:09.04,Default,,0000,0000,0000,,so he basically does.
Dialogue: 0,0:03:09.92,0:03:12.02,Default,,0000,0000,0000,,He knows she needs practice.
Dialogue: 0,0:03:12.04,0:03:13.62,Default,,0000,0000,0000,,He wants to give her control.
Dialogue: 0,0:03:14.25,0:03:17.64,Default,,0000,0000,0000,,But he also knows she’d be slower\Nand make more mistakes,
Dialogue: 0,0:03:17.67,0:03:19.17,Default,,0000,0000,0000,,and his patient comes first.
Dialogue: 0,0:03:20.25,0:03:24.88,Default,,0000,0000,0000,,So Kristin has no hope of getting anywhere\Nnear those nerves during this rotation.
Dialogue: 0,0:03:25.42,0:03:29.79,Default,,0000,0000,0000,,She’ll be lucky if she operates more than\N15 minutes during a four-hour procedure.
Dialogue: 0,0:03:30.25,0:03:32.88,Default,,0000,0000,0000,,And she knows that when she slips up,
Dialogue: 0,0:03:33.46,0:03:36.50,Default,,0000,0000,0000,,he’ll tap a touch screen,\Nand she’ll be watching again,
Dialogue: 0,0:03:36.92,0:03:39.54,Default,,0000,0000,0000,,feeling like a kid in the corner\Nwith a dunce cap.
Dialogue: 0,0:03:41.58,0:03:45.08,Default,,0000,0000,0000,,Like all the studies of robots and work\NI’ve done in the last eight years,
Dialogue: 0,0:03:45.11,0:03:47.23,Default,,0000,0000,0000,,I started this one\Nwith a big, open question:
Dialogue: 0,0:03:47.25,0:03:50.04,Default,,0000,0000,0000,,How do we learn to work\Nwith intelligent machines?
Dialogue: 0,0:03:50.79,0:03:56.60,Default,,0000,0000,0000,,To find out, I spent two and a half years\Nobserving dozens of residents and surgeons
Dialogue: 0,0:03:56.62,0:04:00.10,Default,,0000,0000,0000,,doing traditional and robotic surgery,\Ninterviewing them
Dialogue: 0,0:04:00.12,0:04:03.46,Default,,0000,0000,0000,,and in general hanging out\Nwith the residents as they tried to learn.
Dialogue: 0,0:04:04.25,0:04:07.60,Default,,0000,0000,0000,,I covered 18 of the top\NUS teaching hospitals,
Dialogue: 0,0:04:07.62,0:04:09.08,Default,,0000,0000,0000,,and the story was the same.
Dialogue: 0,0:04:09.88,0:04:12.42,Default,,0000,0000,0000,,Most residents were in Kristen's shoes.
Dialogue: 0,0:04:12.96,0:04:14.75,Default,,0000,0000,0000,,They got to “see one” plenty,
Dialogue: 0,0:04:15.58,0:04:17.88,Default,,0000,0000,0000,,but the “do one” was barely available.
Dialogue: 0,0:04:18.33,0:04:20.86,Default,,0000,0000,0000,,So they couldn’t struggle,\Nand they weren’t learning.
Dialogue: 0,0:04:21.29,0:04:25.10,Default,,0000,0000,0000,,This was important news for surgeons, but\NI needed to know how widespread it was:
Dialogue: 0,0:04:25.12,0:04:28.96,Default,,0000,0000,0000,,Where else was using AI\Nblocking learning on the job?
Dialogue: 0,0:04:30.21,0:04:34.52,Default,,0000,0000,0000,,To find out, I’ve connected with a small\Nbut growing group of young researchers
Dialogue: 0,0:04:34.54,0:04:37.98,Default,,0000,0000,0000,,who’ve done boots-on-the-ground studies\Nof work involving AI
Dialogue: 0,0:04:38.00,0:04:40.98,Default,,0000,0000,0000,,in very diverse settings\Nlike start-ups, policing,
Dialogue: 0,0:04:41.00,0:04:43.60,Default,,0000,0000,0000,,investment banking and online education.
Dialogue: 0,0:04:43.62,0:04:49.48,Default,,0000,0000,0000,,Like me, they spent at least a year\Nand many hundreds of hours observing,
Dialogue: 0,0:04:49.50,0:04:53.42,Default,,0000,0000,0000,,interviewing and often working\Nside-by-side with the people they studied.
Dialogue: 0,0:04:54.46,0:04:56.88,Default,,0000,0000,0000,,We shared data, and I looked for patterns.
Dialogue: 0,0:04:57.92,0:05:03.12,Default,,0000,0000,0000,,No matter the industry, the work,\Nthe AI, the story was the same.
Dialogue: 0,0:05:04.04,0:05:07.68,Default,,0000,0000,0000,,Organizations were trying harder\Nand harder to get results from AI,
Dialogue: 0,0:05:07.71,0:05:11.25,Default,,0000,0000,0000,,and they were peeling learners away from\Nexpert work as they did it.
Dialogue: 0,0:05:12.33,0:05:15.21,Default,,0000,0000,0000,,Start-up managers were outsourcing\Ntheir customer contact.
Dialogue: 0,0:05:15.83,0:05:19.88,Default,,0000,0000,0000,,Cops had to learn to deal with crime\Nforecasts without experts support.
Dialogue: 0,0:05:20.88,0:05:24.12,Default,,0000,0000,0000,,Junior bankers were getting\Ncut out of complex analysis,
Dialogue: 0,0:05:24.50,0:05:27.58,Default,,0000,0000,0000,,and professors had to build\Nonline courses without help.
Dialogue: 0,0:05:29.12,0:05:32.35,Default,,0000,0000,0000,,And the effect of all of this\Nwas the same as in surgery.
Dialogue: 0,0:05:32.38,0:05:35.29,Default,,0000,0000,0000,,Learning on the job\Nwas getting much harder.
Dialogue: 0,0:05:36.96,0:05:38.38,Default,,0000,0000,0000,,This can’t last.
Dialogue: 0,0:05:39.54,0:05:43.81,Default,,0000,0000,0000,,McKinsey estimates that between half\Na billion and a billion of us
Dialogue: 0,0:05:43.83,0:05:47.96,Default,,0000,0000,0000,,are going to have to adapt to AI\Nin our daily work by 2030.
Dialogue: 0,0:05:49.00,0:05:51.01,Default,,0000,0000,0000,,And we’re assuming\Nthat on-the-job learning
Dialogue: 0,0:05:51.04,0:05:52.95,Default,,0000,0000,0000,,will be there for us as we try.
Dialogue: 0,0:05:53.50,0:05:57.77,Default,,0000,0000,0000,,Accenture’s latest workers survey showed\Nthat most workers learned key skills
Dialogue: 0,0:05:57.79,0:06:00.08,Default,,0000,0000,0000,,on the job, not in formal training.
Dialogue: 0,0:06:01.29,0:06:04.81,Default,,0000,0000,0000,,So while we talk a lot about its\Npotential future impact,
Dialogue: 0,0:06:04.83,0:06:08.52,Default,,0000,0000,0000,,the aspect of AI\Nthat may matter most right now
Dialogue: 0,0:06:08.54,0:06:11.92,Default,,0000,0000,0000,,is that we’re handling it in a way\Nthat blocks learning on the job
Dialogue: 0,0:06:12.38,0:06:14.00,Default,,0000,0000,0000,,just when we need it most.
Dialogue: 0,0:06:15.46,0:06:21.50,Default,,0000,0000,0000,,Now across all our sites,\Na small minority found a way to learn.
Dialogue: 0,0:06:23.62,0:06:26.67,Default,,0000,0000,0000,,They did it by breaking and bending rules.
Dialogue: 0,0:06:27.08,0:06:31.73,Default,,0000,0000,0000,,Approved methods weren’t working,\Nso they bent and broke rules
Dialogue: 0,0:06:31.75,0:06:33.73,Default,,0000,0000,0000,,to get hands-on practice with experts.
Dialogue: 0,0:06:33.75,0:06:39.35,Default,,0000,0000,0000,,In my setting, residents got involved\Nin robotic surgery in medical school
Dialogue: 0,0:06:39.38,0:06:42.96,Default,,0000,0000,0000,,at the expense\Nof their generalist education.
Dialogue: 0,0:06:44.42,0:06:50.27,Default,,0000,0000,0000,,And they spent hundreds of extra hours\Nwith simulators and recordings of surgery,
Dialogue: 0,0:06:50.29,0:06:52.83,Default,,0000,0000,0000,,when you were supposed to learn in the OR.
Dialogue: 0,0:06:53.38,0:06:56.85,Default,,0000,0000,0000,,And maybe most importantly,\Nthey found ways to struggle
Dialogue: 0,0:06:56.88,0:07:00.62,Default,,0000,0000,0000,,in live procedures\Nwith limited expert supervision.
Dialogue: 0,0:07:01.79,0:07:06.10,Default,,0000,0000,0000,,I call all this “shadow learning,”\Nbecause it bends the rules
Dialogue: 0,0:07:06.12,0:07:08.12,Default,,0000,0000,0000,,and learner’s do it out of the limelight.
Dialogue: 0,0:07:09.54,0:07:13.64,Default,,0000,0000,0000,,And everyone turns a blind eye\Nbecause it gets results.
Dialogue: 0,0:07:13.67,0:07:16.83,Default,,0000,0000,0000,,Remember, these are\Nthe star pupils of the bunch.
Dialogue: 0,0:07:17.79,0:07:21.00,Default,,0000,0000,0000,,Now, obviously, this is not OK,\Nand it’s not sustainable.
Dialogue: 0,0:07:21.71,0:07:23.89,Default,,0000,0000,0000,,No one should have to risk getting fired
Dialogue: 0,0:07:23.92,0:07:26.07,Default,,0000,0000,0000,,to learn the skills\Nthey need to do their job.
Dialogue: 0,0:07:26.79,0:07:28.85,Default,,0000,0000,0000,,But we do need to learn from these people.
Dialogue: 0,0:07:29.92,0:07:32.17,Default,,0000,0000,0000,,They took serious risks to learn.
Dialogue: 0,0:07:32.79,0:07:37.14,Default,,0000,0000,0000,,They understood they needed to protect\Nstruggle and challenge in their work
Dialogue: 0,0:07:37.17,0:07:40.06,Default,,0000,0000,0000,,so that they could push themselves\Nto tackle hard problems
Dialogue: 0,0:07:40.08,0:07:42.04,Default,,0000,0000,0000,,right near the edge of their capacity.
Dialogue: 0,0:07:42.46,0:07:44.67,Default,,0000,0000,0000,,They also made sure\Nthere was an expert nearby
Dialogue: 0,0:07:44.70,0:07:47.79,Default,,0000,0000,0000,,to offer pointers and to backstop\Nagainst catastrophe.
Dialogue: 0,0:07:48.88,0:07:52.33,Default,,0000,0000,0000,,Let’s build this combination\Nof struggle and expert support
Dialogue: 0,0:07:52.71,0:07:55.46,Default,,0000,0000,0000,,into each AI implementation.
Dialogue: 0,0:07:56.38,0:07:59.20,Default,,0000,0000,0000,,Here’s one clear example\NI could get of this on the ground.
Dialogue: 0,0:08:00.12,0:08:01.35,Default,,0000,0000,0000,,Before robots,
Dialogue: 0,0:08:01.38,0:08:06.17,Default,,0000,0000,0000,,if you were a bomb disposal technician,\Nyou dealt with an IED by walking up to it.
Dialogue: 0,0:08:07.33,0:08:09.48,Default,,0000,0000,0000,,A junior officer was\Nhundreds of feet away,
Dialogue: 0,0:08:09.50,0:08:12.81,Default,,0000,0000,0000,,so could only watch and help\Nif you decided it was safe
Dialogue: 0,0:08:12.83,0:08:14.25,Default,,0000,0000,0000,,and invited them downrange.
Dialogue: 0,0:08:15.21,0:08:19.10,Default,,0000,0000,0000,,Now you sit side-by-side\Nin a bomb-proof truck.
Dialogue: 0,0:08:19.12,0:08:20.93,Default,,0000,0000,0000,,You both watched the video feed.
Dialogue: 0,0:08:20.96,0:08:25.27,Default,,0000,0000,0000,,They control a distant robot,\Nand you guide the work out loud.
Dialogue: 0,0:08:25.29,0:08:28.50,Default,,0000,0000,0000,,Trainees learn better than they\Ndid before robots.
Dialogue: 0,0:08:29.12,0:08:33.06,Default,,0000,0000,0000,,We can scale this to surgery,\Nstart-ups, policing,
Dialogue: 0,0:08:33.08,0:08:35.71,Default,,0000,0000,0000,,investment banking,\Nonline education and beyond.
Dialogue: 0,0:08:36.38,0:08:38.88,Default,,0000,0000,0000,,The good news is\Nwe’ve got new tools to do it.
Dialogue: 0,0:08:39.75,0:08:43.83,Default,,0000,0000,0000,,The internet and the cloud mean we don’t\Nalways need one expert for every trainee,
Dialogue: 0,0:08:44.17,0:08:48.62,Default,,0000,0000,0000,,for them to be physically near each other\Nor even to be in the same organization.
Dialogue: 0,0:08:49.29,0:08:52.33,Default,,0000,0000,0000,,And we can build AI to help:
Dialogue: 0,0:08:53.17,0:08:58.23,Default,,0000,0000,0000,,to coach learners as they struggle,\Nto coach experts as they coach
Dialogue: 0,0:08:58.25,0:09:00.79,Default,,0000,0000,0000,,and to connect those two groups\Nin smart ways.
Dialogue: 0,0:09:03.38,0:09:05.92,Default,,0000,0000,0000,,There are people at work\Non systems like this,
Dialogue: 0,0:09:06.33,0:09:09.12,Default,,0000,0000,0000,,but they’ve been mostly focused\Non formal training.
Dialogue: 0,0:09:09.46,0:09:12.04,Default,,0000,0000,0000,,And the deeper crisis\Nis in on-the-job learning.
Dialogue: 0,0:09:12.42,0:09:14.27,Default,,0000,0000,0000,,We must do better.
Dialogue: 0,0:09:14.29,0:09:16.88,Default,,0000,0000,0000,,Today’s problems demand we do better
Dialogue: 0,0:09:17.38,0:09:22.25,Default,,0000,0000,0000,,to create work that takes full advantage\Nof AI’s amazing capabilities
Dialogue: 0,0:09:23.04,0:09:25.79,Default,,0000,0000,0000,,while enhancing our skills as we do it.
Dialogue: 0,0:09:26.33,0:09:29.08,Default,,0000,0000,0000,,That’s the kind of future\NI dreamed of as a kid.
Dialogue: 0,0:09:29.46,0:09:31.62,Default,,0000,0000,0000,,And the time to create it is now.
Dialogue: 0,0:09:32.33,0:09:33.56,Default,,0000,0000,0000,,Thank you.
Dialogue: 0,0:09:33.58,0:09:37.21,Default,,0000,0000,0000,,(Applause)