The New Yorker - USA (2019-09-30)

THENEWYORKER,SEPTEMBER30, 2019 31

cause I’m touching a business of billions
around the world.”
We walked over to another empty
space. This was where he planned to
build a remote “cockpit” for surgeries in
which the patient was not in the same
room. It was the original darpa proj-
ect, reborn. “I think with 5G coming we
can do it,” Giulianotti said. He will have
to work quickly: a Boston-area company
called Vicarious Surgical, which is partly
funded by Bill Gates, is also working on
a robot that a surgeon will be able to
operate from a remote distance.
Even though competitors like
Vicarious Surgical are beginning to
emerge, Intuitive Surgical retains an
overwhelming market share, and, with
five billion dollars in cash reserves, it
can afford to invest heavily in R. & D.
An important next step for the com-
pany will be upgrading the da Vinci.
With its cautionary beeps, the robot is
more than a tool, but it is hardly as au-
tonomous as, say, a self-driving car. It
marshals no personal data about the
patient or the population that has the
same conditions; it does not make as-
sessments by weighing genetic infor-
mation or by aggregating data from

similar procedures. During an opera-

tion, a da Vinci offers a surgeon only

rudimentary guidance. If its software

were a chess program, all it would do

is keep you from accidentally sacrific-

ing your queen on the next move.

Giulianotti told me that a more ad-

vanced robot could have assessed the

tumor he saw that morning better in

situ than he could: “I’m pretty sure that

the computer would be able to recog-

nize—based on the pattern of blood

flow and the tissue itself, and based on

billions of people—what is the best de-

cision: ‘You can save the spleen,’ or ‘It’s

better to remove the spleen.’” Later, I

learned that the patient’s tumor was not

actually malignant; she could have kept

her spleen.

A company called Digital Surgery

is trying to smarten robots by feeding

visual data sets of surgical procedures

into artificial-intelligence algorithms.

The company already markets an app

that trains doctors through simulated

surgery, and it is essentially applying the

same technique to training robots. The

company’s founder, a surgeon named

Jean Nehme, told me, “We’re not any-

where near playing grand-master chess.

But the computers are at the level of a

medical-school student. Our algorithms

recognize and understand where a sur-

geon is in a procedure.”

Fred Moll, the Intuitive Surgical co-

founder, is eager to see robotic medical

devices incorporate artificial intelligence,

but he argues that there are some de-

cisions a computer simply can’t make.

He asked me to imagine a surgeon re-

moving a tumor from a patient’s brain.

Too much cutting could lead to a loss

of function, such as aphasia; too little

cutting could leave the patient open to

a possibly fatal outcome. The patient,

meanwhile, is awake on the operating

table, providing the surgeon with sec-

ond-by-second feedback. “You’re trying

to make a judgment about how much

should I take, and there’s patient inter-

action,” Moll said. “When do you stop?

There’s a component that’s going to be

hard to displace onto a robot.”

Intuitive Surgical tends to point to

the F.D.A. as the reason that complex

artificial intelligence hasn’t yet made it

into the operating room. Last year,

Myriam Curet, the company’s chief

medical officer, spoke to the Robot Re-

port, a Web site, and said, “I actually

think the technology to create an au-

tonomous robot will actually be easy to

solve.... The problem will be the reg-

ulatory environment.” Consumer fears

will also have to be overcome. Gary Gu-

thart, the C.E.O. of Intuitive Surgical,

reminded me that human pilots still

take off and land commercial planes,

even though they don’t have to do so.

He said, “Flight-wise, I think most

folks, while they accept that there’s a

fair amount of automation, they want

the pilot in there. They want Sully Sul-

lenberger.” He was careful not to prom-

ise too much autonomy for the da Vinci

too soon. “When the computer makes

a recommendation, it better be right,”

he said. For the moment, Intuitive Sur-

gical seems focussed more on making

humans as good as robots than on the

reverse. Intuitive’s main automation goal,

Curet explained, is to dampen the vari-

ability of a human surgeon’s perfor-

mance—“‘My child was throwing up,

so therefore I’m tired today, and there-

fore my hands are not as steady as they

were yesterday.’”

Some of Intuitive’s key patents re-

lated to the da Vinci have expired or

oneself, of decadent uselessness,

though one might, of course, always alarm

to some emergency, a child caught

in the undertow, say, who must be

dragged to shore and breathed into

like an empty balloon, an empty balloon

on which everything depends, might,

bent over the small body, waiting for it

to rise, to float, casting a shadow

the size of oneself, not even see,

though one was, of course, warned

it would come, and soon, the shadow

of that wave, like a new sky, already

overhead and even now descending.

—Annelyse Gelman