WEBVTT

00:00:00.000 --> 00:00:04.234
One executive said the work should move ahead with far more care.

00:00:04.234 --> 00:00:08.469
A head of state brushed that aside within hours.

00:00:08.469 --> 00:00:12.703
The quarrel never stayed in the lab; it landed at the centre of power.

00:00:12.703 --> 00:00:16.938
How far this technology may be allowed to run has stopped being a matter for

00:00:16.938 --> 00:00:21.172
researchers alone.

00:00:21.172 --> 00:00:25.407
It reaches the people who make medicines too.

00:00:25.407 --> 00:00:29.641
Somebody has to vouch for the tools used to build a drug, and it is no longer clear

00:00:29.641 --> 00:00:33.876
who does that, or on what basis.

00:00:33.876 --> 00:00:38.110
What happened today pushed that problem into the open all at once.

00:00:38.110 --> 00:00:42.428
One executive asked his industry to proceed with more care.

00:00:42.428 --> 00:00:46.745
Within a day the country's leader dismissed him by name.

00:00:46.745 --> 00:00:51.062
An argument like this normally travels slowly.

00:00:51.062 --> 00:00:55.380
It passes through committees and draft bills first.

00:00:55.380 --> 00:00:59.697
This time all of that was skipped.

00:00:59.697 --> 00:01:04.015
The speed of the reply says more than the words did.

00:01:04.015 --> 00:01:08.332
It tells you how far the stakes have grown.Two camps line up here, and both sit

00:01:08.332 --> 00:01:12.650
inside the same business.

00:01:12.650 --> 00:01:16.968
The break did not come from outside; it opened within.

00:01:16.968 --> 00:01:21.285
People who sell the same technology now say opposite things about how dangerous it

00:01:21.285 --> 00:01:25.602
is.

00:01:25.602 --> 00:01:29.920
For a buyer that is a serious position.

00:01:29.920 --> 00:01:34.237
One seller calls its own product perilous while another calls it fine.

00:01:34.237 --> 00:01:38.555
Deciding whom to believe is now separate from judging the technology.Choosing to go

00:01:38.555 --> 00:01:42.873
more gently is not a new idea.

00:01:42.873 --> 00:01:47.190
What changed is that the idea turned into a banner.

00:01:47.190 --> 00:01:51.507
Once it becomes a flag, people argue about sides rather than substance.

00:01:51.507 --> 00:01:55.825
Whether restraint is actually warranted then gets pushed to the back of the queue.

00:01:55.825 --> 00:02:00.142
When a technical question turns into a party flag, standing comes before substance.

00:02:00.142 --> 00:02:04.460
With no single law in sight, companies moved to draw their own lines.

00:02:04.460 --> 00:02:08.440
With the politicians divided, no single law is going to arrive soon.

00:02:08.440 --> 00:02:12.420
That is why companies began drawing their own boundaries.

00:02:12.420 --> 00:02:16.400
Rules may come later, but nobody can predict what they will say.

00:02:16.400 --> 00:02:20.380
Draw the line yourself first and you at least choose what it says.

00:02:20.380 --> 00:02:24.360
Moving early pays.

00:02:24.360 --> 00:02:28.340
The price of waiting is losing the pen.Three moves landed on a single day: a written

00:02:28.340 --> 00:02:32.320
rulebook, a shared institution, and a call for a hard stop.

00:02:32.320 --> 00:02:36.300
None of them compels anyone.

00:02:36.300 --> 00:02:40.280
Each rests on firms keeping promises they wrote for themselves, and breaking one

00:02:40.280 --> 00:02:44.260
carries no penalty.

00:02:44.260 --> 00:02:48.240
It does carry a cost in reputation.

00:02:48.240 --> 00:02:52.220
The force here comes from the market, not from law.The machinery for stopping a

00:02:52.220 --> 00:02:56.200
system on the spot is simple enough.

00:02:56.200 --> 00:03:00.180
The hard part is who may pull it, and when.

00:03:00.180 --> 00:03:04.160
Until that is settled, the mechanism sits there and does nothing.

00:03:04.160 --> 00:03:08.140
Pull too late and it saves nobody; too early and you halt sound work.

00:03:08.140 --> 00:03:12.120
If such a thing sat inside a tool for making medicines, who presses it?

00:03:12.120 --> 00:03:16.100
Whether that promise holds is about to be tested in public.

00:03:16.100 --> 00:03:20.080
One early test was what traders did next.

00:03:20.090 --> 00:03:24.122
When politics and business pull apart, the first thing to move is money.

00:03:24.122 --> 00:03:28.154
Traders act long before anyone writes an analysis.

00:03:28.154 --> 00:03:32.185
A day after the text appeared, the chipmakers fell.

00:03:32.185 --> 00:03:36.217
Nothing about the technology itself had changed.

00:03:36.217 --> 00:03:40.249
Only the mood had.

00:03:40.249 --> 00:03:44.281
Funding here now tracks the political weather, not the work.

00:03:44.281 --> 00:03:48.313
That is structural, not a passing dip.One document, one public rebuke, and the

00:03:48.313 --> 00:03:52.345
semiconductor names took the hit.

00:03:52.345 --> 00:03:56.376
The trouble is how short that path is.

00:03:56.376 --> 00:04:00.408
The worth of the underlying work never moved, yet capital did.

00:04:00.408 --> 00:04:04.440
Those same components run the computing behind new medicines.

00:04:04.440 --> 00:04:08.472
If their makers wobble with politics, supply itself starts to look uncertain.

00:04:08.472 --> 00:04:12.504
Plans built on steady supply rest on the mood of a week.A single piece of writing

00:04:12.504 --> 00:04:16.535
moved a market.

00:04:16.535 --> 00:04:20.567
Not its reasoning, but its author.

00:04:20.567 --> 00:04:24.599
Had an unknown researcher put out the identical text, nothing would have happened.

00:04:24.599 --> 00:04:28.631
The sums that shifted show what a job title is worth.

00:04:28.631 --> 00:04:32.663
The market responded to the person, not the argument.

00:04:32.663 --> 00:04:36.695
Debate about the technology is being swallowed by rank.

00:04:36.695 --> 00:04:40.726
Doubt about supply, and about where capital goes, lands on the people who buy these

00:04:40.726 --> 00:04:44.758
tools.

00:04:44.758 --> 00:04:48.790
One of those buyers is the medicine business.

00:04:48.790 --> 00:04:53.529
While the row spreads, capital keeps arriving for tools that catch harm early.

00:04:53.529 --> 00:04:58.269
Holding back and pressing ahead run together, and that is no contradiction.

00:04:58.269 --> 00:05:03.008
The quarrel raises concern about harm, and that concern pulls in capital.

00:05:03.008 --> 00:05:07.748
The argument itself works as a tailwind here.Three things at once: a tool that spots

00:05:07.748 --> 00:05:12.487
harm, backing for the opening stages of research, and support for the local

00:05:12.487 --> 00:05:17.227
language.

00:05:17.227 --> 00:05:21.966
What they share is a direction.

00:05:21.966 --> 00:05:26.706
Capital is moving to the front of the process, not the back.

00:05:26.706 --> 00:05:31.445
Rather than catching trouble once a medicine reaches patients, the aim is to prevent

00:05:31.445 --> 00:05:36.185
it while the work is under way.

00:05:36.185 --> 00:05:40.924
Set against recalls and lawsuits, that is the cheaper bet.The sum going into early

00:05:40.924 --> 00:05:45.664
detection of harm is not large by the standards of this field.

00:05:45.664 --> 00:05:50.403
What matters is that dedicated capital arrived at all.

00:05:50.403 --> 00:05:55.143
Protective technology used to sit in the books as an expense.

00:05:55.143 --> 00:05:59.882
It is starting to be treated as an investment.

00:05:59.882 --> 00:06:04.622
Expenses get cut; investments get grown.

00:06:04.622 --> 00:06:09.361
That change of framing rewrites how budgets are built.

00:06:09.361 --> 00:06:14.101
Capital gathering around protective technology is a tailwind.

00:06:14.101 --> 00:06:18.840
Sorting out the law around using those tools has barely begun.

00:06:18.840 --> 00:06:22.869
The decision to bring in a technology never ends with performance.

00:06:22.869 --> 00:06:26.897
If nobody knows who answers for the outcome, no organisation can adopt it, however

00:06:26.897 --> 00:06:30.926
capable it is.

00:06:30.926 --> 00:06:34.955
Better engineering will not clear that barrier alone.

00:06:34.955 --> 00:06:38.983
Institutions have to catch up.

00:06:38.983 --> 00:06:43.012
Until they do, people work by feel.

00:06:43.012 --> 00:06:47.040
Look at what that involves.One group of specialists set out the open questions.

00:06:47.040 --> 00:06:51.069
Who owns an invention.

00:06:51.069 --> 00:06:55.098
How regulators treat findings made by a machine.

00:06:55.098 --> 00:06:59.126
Who carries the blame for a bad call.

00:06:59.126 --> 00:07:03.155
None of these is technical.

00:07:03.155 --> 00:07:07.184
Each asks how a company is run.

00:07:07.184 --> 00:07:11.212
However strong the tool, adoption halts where nobody can answer them.

00:07:11.212 --> 00:07:15.241
Where those answers exist, they become an edge rather than a wall.Proving something

00:07:15.241 --> 00:07:19.270
behaves as expected, and keeping the evidence, is already compulsory in medicine.

00:07:19.270 --> 00:07:23.298
Whether the old method fits a tool that learns and shifts is unsettled.

00:07:23.298 --> 00:07:27.327
A fixed mechanism you check once.

00:07:27.327 --> 00:07:31.355
A learning one may act differently at the hour of use than at the check.

00:07:31.355 --> 00:07:35.384
That is where the established approach runs out.

00:07:35.384 --> 00:07:39.413
Designing what replaces it is today's problem.

00:07:39.413 --> 00:07:43.441
As the legal picture firms up, the case for bringing a tool in gets clearer.

00:07:43.441 --> 00:07:47.470
Next, how all this shifts the way tools are picked on the ground.

00:07:47.470 --> 00:07:52.126
Take everything so far from the seat of someone working in this industry.

00:07:52.126 --> 00:07:56.782
A fight in politics, promises written in private, a market reaction, capital in

00:07:56.782 --> 00:08:01.438
motion, questions of law.

00:08:01.438 --> 00:08:06.094
All of it is happening at once, and it makes the choice of what to bring in harder

00:08:06.094 --> 00:08:10.751
than it used to be.

00:08:10.751 --> 00:08:15.407
Picking on capability and price alone is ending.

00:08:15.407 --> 00:08:20.063
There is more to check now: how a vendor is run inside, how ready it is for the law,

00:08:20.063 --> 00:08:24.719
even how the thing gets switched off.How a vendor guards what it builds is becoming

00:08:24.719 --> 00:08:29.375
a yardstick beside capability and price.

00:08:29.375 --> 00:08:34.031
Not only what a system can do, but how it is kept in hand.

00:08:34.031 --> 00:08:38.687
At the contract stage you have to establish how it halts and how it gets tested.

00:08:38.687 --> 00:08:43.343
Comparing vendors takes more than clauses; it takes operating records.

00:08:43.343 --> 00:08:47.999
Without those records, the arrangement lives only on paper.

00:08:47.999 --> 00:08:52.656
Vendors who can show a track record will win the work.

00:08:52.656 --> 00:08:57.312
Judging that takes time, and skipping it is the kind of saving you cannot undo.

00:08:57.312 --> 00:09:01.968
If the basis for choosing changes, contracts and testing routines need review too.

00:09:01.968 --> 00:09:06.624
Whether you can start that work is the practical question this day leaves you.

00:09:06.624 --> 00:09:11.280
But some conditions never made it into the reporting.

00:09:11.280 --> 00:09:15.183
Reporters told the story of a fight.

00:09:15.183 --> 00:09:19.086
They left some conditions out.

00:09:19.086 --> 00:09:22.990
Judge from what surfaced alone and an unseen assumption trips you.

00:09:22.990 --> 00:09:26.893
Acting only on visible information is risky.

00:09:26.893 --> 00:09:30.796
What matters is not who said what, but what moves underneath the words.

00:09:30.796 --> 00:09:34.699
Behind every article we get to read sits something that never became an

00:09:34.699 --> 00:09:38.602
article.Three points stand out.

00:09:38.602 --> 00:09:42.505
People preach caution while their own rollouts speed up.

00:09:42.505 --> 00:09:46.409
A large state has pushed back against the whole thing.

00:09:46.409 --> 00:09:50.312
And the joint institution has no substance yet.

00:09:50.312 --> 00:09:54.215
The gap between what is said and what is done is the defining feature here.

00:09:54.215 --> 00:09:58.118
Firms urge restraint and sign more deals in the same breath.

00:09:58.118 --> 00:10:02.021
Miss that contradiction and you will be steered by announcements.

00:10:02.021 --> 00:10:05.925
The pushback from that state shows the danger that restraint ends up running on one

00:10:05.925 --> 00:10:09.828
side only.

00:10:09.828 --> 00:10:13.731
Whoever halts alone falls behind.

00:10:13.731 --> 00:10:17.634
While that holds, voluntary promises will always meet competitive pressure.

00:10:17.634 --> 00:10:21.537
Doing nothing, though, is worse again.

00:10:21.537 --> 00:10:25.440
All you can do is hold words and deeds side by side and compare them coolly.

00:10:25.440 --> 00:10:29.344
The gap between talk and deeds makes judgment hard.

00:10:29.344 --> 00:10:33.247
Looking past the announcements to what is actually moving is now part of the job for

00:10:33.247 --> 00:10:37.150
whoever picks the tools.

00:10:37.150 --> 00:10:41.216
Today a question about limits became a matter for power.

00:10:41.216 --> 00:10:45.281
Companies reached to set their own boundaries, the market answered, and capital went

00:10:45.281 --> 00:10:49.347
toward protection.

00:10:49.347 --> 00:10:53.413
The gap between saying and doing is still open.

00:10:53.413 --> 00:10:57.479
So for people who make medicines, the question is shifting from which tool to use to

00:10:57.479 --> 00:11:01.544
whose organisation you are willing to depend on.

00:11:01.544 --> 00:11:05.610
Tomorrow asks whether those arrangements have anything inside them.
