Last week the Senate confirmed Stephanie O’Sullivan to be the Principal Deputy Director of National Intelligence (PDDNI), the number two position in the Office of the Director of National Intelligence (ODNI). Ms. O’Sullivan, a longtime CIA employee who is currently Associate Deputy Director of CIA, will be the fourth person to hold that office.
A review of the materials submitted in support of her nomination, especially a set of answers to pre-hearing questions (pdf) submitted by the Senate Intelligence Committee, turns up a number of interesting new details concerning intelligence policy. For example:
** ODNI began an effort last year “to reinvigorate the declassification of imagery for public release,” Ms. O’Sullivan said. She indicated that this program had been “launched in May 2010,” but there is little sign that it has had any impact to date.
On the contrary, attempts to gain access to historical intelligence satellite imagery remain as contentious and mostly as fruitless as ever. “Trying to get [satellite imagery] declassified today, it’s like I’m taking their marbles away,” said Charles P. Vick of GlobalSecurity.org earlier this month, “and it’s over 40 years old.” (“New Secrets of Huge Soviet Moon Rocket Revealed” by Leonard David, Space.com, February 7, 2011).
** Over the past several years, science and technology research activities at CIA “have improved substantially,” leading to the operational use of new intelligence technologies, Ms. O’Sullivan told the Senate Intelligence Committee.
This upbeat assertion contrasts, for example, with a 2006 Intelligence Science Board report which found that technological “innovation in the Intelligence Community” was “hobble[d]” and that the problem was approaching “critical levels.” (“Intel Science Board on ‘The New S&T Landscape’,” Secrecy News, January 10, 2011).
But Ms. O’Sullivan said that “There has been a marked increase in cutting edge technology which has been developed through research and development and successfully transitioned to operations… [The CIA Directorate of Science and Technology’s] acquisitions and operations have also delivered innovative new capabilities and technical operations that have closed collection gaps.”
The nature of these capabilities and operations was not specified. One trusts that they did not include the fraudulent intelligence-related technologies that were central to the unfolding Dennis Montgomery/e-Treppid Technologies scandal and featured in the New York Times last weekend. (“Government Tries to Keep Secret What Many Consider a Fraud” by Eric Lichtblau and James Risen, New York Times, February 20, 2011).
Ms. O’Sullivan also provided brief commentary on the obscure and mostly classified Comprehensive National Cybersecurity Initiative (CNCI), which she said was “progressing satisfactorily.” She discussed in general terms the government’s response to WikiLeaks, which she said “represents the magnitude of the potential danger when technology, opportunity, and malevolent human motivation combine.” Her answers to the pre-nomination hearing questions may be found here.
Sometimes the questions that were posed to Ms. O’Sullivan were more interesting than the answers. Thus, in an oddly flattering formulation, the Senate Intelligence Committee observed that “The National Reconnaissance Office has historically attempted to deliver more program content that can be accomplished under the budgets requested by the President or appropriated by Congress. We understand that this problem is again arising in the fiscal year 2012 budget build now in progress.”
A routine questionnaire for Presidential nominees asks whether the nominee has ever been arrested or charged with committing a crime. Ms. O’Sullivan noted (pdf) that she was charged with trespassing in August 1989 for “being in a public park after closing.” The charges were subsequently dropped.
Commercial artificial intelligence tools have recently emerged that are able to produce police reports. If the resulting reports are inaccurate, incomplete or biased, or if the process leaks confidential information, this could undermine the criminal justice system and harm citizens.
Too often, affected patients, clinicians, and regulators cannot see how the system works, why a decision was made, or whether meaningful human oversight occurred.
Existing tools from other domains, such as existing robust public engagement processes in drug development, when applied to AI deployment can help strengthen public trust in these systems and enhance perceptions of their legitimacy and the decisions they produce.
With thoughtful policy action, it is still possible to build systems that are fair, transparent, and accountable, and to earn the public trust that will ultimately determine AI’s future. We hope policymakers are ready to act.