Could Artificial Intelligence Be a Big Win For Taxpayers?

Advocates believe it can speed up government mandated processes at the local, state and federal level, improve the efficiency of the judicial system - and possibly generate as much new value as the internet did. JL

Daniel Malloy reports in Ozy:

Imagine government bots designed to slash red tape in patent applications and veterans’ benefits claims. They can save lawyers’ time by sifting through mountains of case law or investigators’ eyes in scanning photos for a fugitive. Artificial intelligence could save the feds up to $41.1 billion per year by freeing up 1.2 billion working hours. (And) A.I. is out to replace tasks, not necessarily jobs, redirecting people to higher-value work.

Imagine government bots designed to slash red tape in patent applications and veterans’ benefits claims. They can save lawyers’ time by sifting through mountains of case law in search of relevant precedent or investigators’ eyes in scanning photos for a fugitive. The artificial intelligence hurricane heading for the economy is also going to hit government, and a study this year by the consultancy firm Deloitte predicts it could save the feds up to $41.1 billion per year by freeing up 1.2 billion working hours.

Add in the impact on state and local governments — could we finally be on the verge of an efficient DMV experience? — and the possibilities multiply. But government is naturally resistant to change, and the Donald Trump administration is sending mixed signals about how much it plans to embrace the bots.

For an administration keen on agency cuts, the prospect of major savings via A.I. should be tantalizing.

In June the president gathered 18 tech industry CEOs at long tables arranged in a rectangle in the State Dining Room of the White House and pledged more than $1 trillion in savings for taxpayers over a decade. “We’re embracing big change, bold thinking and outsider perspectives to transform government and make it the way it should be — and at far less cost,” Trump said, with Apple CEO Tim Cook seated to his right. One seat over from Cook was Jared Kushner, the president’s senior adviser and son-in-law who is tasked with leading the transformation.

But Trump’s public remarks didn’t even mention A.I., and his cabinet secretaries have been skeptical of the talk in the tech world. In March Treasury Secretary Steven Mnuchin told the digital news outlet Axios that he doesn’t expect big job losses from A.I. in the broader economy for 50 to 100 years. “It’s not even on our radar screen,” he said. Experts in the field were aghast. Rishon Blumberg, co-founder of 10X Management, a talent agency for tech professionals, tells OZY the comment was “absolutely absurd” and the administration must change course quickly on retooling the education system and admitting more high-skilled immigrants into the country. “Between education, immigration and sticking your head in the sand regarding what A.I. and machine learning are going to do to the population at large, you’ve got sort of a perfect storm for real problems to occur over the course of the next 25 years,” Blumberg says.

For an administration keen on agency cuts, the prospect of major savings via A.I. should be tantalizing. It would also be more popular across party lines than slashing domestic programs, as the Trump administration has proposed in order to offset military spending hikes. “The question isn’t so much what jobs go away, but how that job changes,” says Ben Pring, co-author of What to Do When Machines Do Everything. “But if your job is literally just tick this box, tick that box, make sure this form is filled in correctly, this is the work that’s going to be susceptible to smart software.”

A.I. is out to replace tasks, not necessarily jobs, redirecting people to higher-value work (and anyway, with civil service protections it’s hard to fire government employees). Take instant translation software: It could speed up real-time language conversion and enable faster distribution of transcripts to the media. But human translators would still be necessary to understand the remarks’ cultural context and make sure diplomacy runs smoothly. At the local level, a public school teacher could use bots to assign and grade homework, allowing more time and energy for student interaction.

Some federal agencies already are deploying these technologies, as the Deloitte report observes. The U.S. Citizenship and Immigration Services has a customer service chatbot named Emma. The Centers for Disease Control and Prevention uses A.I. to track polio. Just as government research funding was instrumental in creating the internet, it also helped push forward the early development of A.I. Futurist Amy Webb points out that some of the biggest government investment in A.I. research was during the Cold War, when the Defense Department wanted to deploy Russian translation technology and plan for doomsday scenarios. Webb, the author of The Signals are Talking: Why Today’s Fringe is Tomorrow’s Mainstream, says the difference now is that A.I. has much broader applications than defense. “This is a critical transition period that is very analogous to the late 1980s, when the internet transitioned away from being an academic outpost to the commercial internet we know today,” Webb says. “That did change a lot about life.”

But just as government can be painfully slow to process your disability benefits application, so too is it slow to take up new technologies. Some Defense Department computers still use 8-inch floppy disks or Windows 95 browsers. Deliberation does have its advantages, as not all A.I. deployments have been gleaming successes: Predictive criminal sentencing algorithms were more likely to falsely label Black defendants as future offenders, according to a ProPublica investigation.

Even as it rolls out robots, machine learning, computer vision and other tools in fits and starts, the government is sure to lag behind private industry in embracing A.I. But as these technologies increasingly power our on-demand lives, voters likely will find the endless DMV line or a six- to-eight-week wait for a new passport unacceptable — and hold officeholders responsible.