Is artificial intelligence a threat? Experts weigh the risks
July 02, 2015
As several tech luminaries express worries about the future of AI, researchers met at a D.C. think tank to discuss whether advances could pose a threat.
By LaWanda Wells and Peter L. Levin, U.S. Department of Veterans Affairs
The laws that govern Federal procurement are carefully designed to be equitable to vendors, transparent to taxpayers, and auditable for mistakes or malfeasance. They are not optimized for speed. There are exceptions to the rules, but these are the rules. And generally they work pretty well.
Of course in the real world blind justice doesn’t make purchasing decisions, people do. And government’s sense of the possible is directly informed by thought leaders from industry and the non-profit sector who reach out to us. These are the people who have spent their entire careers untangling tough problems, and we need – even rely – on their input to better understand our spectrum of choice.
That said, “market research” conversations are occasionally, and unnecessarily, difficult. Sales and marketing professionals sometimes fall victim to the well-remunerated temptation of trying to move dollars in their employers’ direction, instead of helping us better understand what the questions should be. The tension can be palpable.
More importantly, these meetings don’t help anyone get their work done. Vendors who approach these discussions as ordinary sales calls are missing the point, and squandering a great opportunity to teach. Because we are trying to solve old problems in new ways, it would be much better if we invested the time discussing the art of the possible: the cutting edge of performance, of capability, of customer satisfaction, of taxpayer value. Help the public sector better understand best industry practices (not just yours), and how we can demonstrate, pilot, and then measure them.
The most useful discussions we have are with vendors we trust because they behave trustfully. They don’t expect anything except a good talk at the white board. They sincerely complement their competitors, point out areas of substantial overlap, and occasionally mention differentiation, but only if there’s a real difference. They welcome constructive challenge constructively offered.
Business development and sales are related but not synonymous. The former seeks mind-share while the latter pursues market-share.
Earlier this year the White House instructed Federal CIOs to “select IT based on appropriate criteria while analyzing available alternatives including proprietary, open source, and mixed source technologies.” The Administration’s clear guidance - well instanced by VA’s creation of OSEHRA (www.osehra.org), an open source custodial agent for VistA - invites industry to engage Federal customers in a new dialogue that values mind-share more than ever before.
The renewed emphasis on modular, open systems with substitutable applications means more competition, more innovation, better value, and shorter implementation timelines. It also means that inevitable development errors are caught earlier, mistakes more easily (and inexpensively) corrected, and platforms more broadly shared and re-used. What’s not to like?
Besides the flexibility of creating smartphone-like applications inside the vast Federal (and federated) IT infrastructure, this “choice architecture” (in the vernacular of White House Office of Information and Regulatory Affairs administrator Cass Sunstein) widens the aperture of access, and flattens the playing field for competitors. Instead of facing the Damocles sword of choosing once and wisely, Federal program managers and acquisition officials are learning how to continuously improve their outcomes with operational models that escape the captivity of single vendors with long term deals, liberate their data from their calcified silos, and provide better value to taxpayers and stakeholders.
Federal CIOs and CFOs are all too familiar with the procurement hazards of large enterprise technology systems: they are prone to the indignities of massive cost overruns and failed features, or miss the government’s performance requirements. It’s simple to show that the chances of correctly walking a three step path from end-to-end is only about 50% even if each step is 90% accurate.
Most projects have way more than three steps. And most steps have way less than a 90% hit rate. Think about it. We know how and why projects fail. Now we have an opportunity to do something about it.
So trying to convince a public-sector customer that only one solution is the right solution - patently, exclusively, miraculously! - doesn’t address the real question anymore, if it ever did.
Let’s talk instead about measured best value, data liberation, and openly architected UN-bundled apps. This is a great way to deliver IT-enabled services that are responsive to market innovation, and can meet the public’s expectation of timing, cost, transparency, and outcomes.