Big Brother is Watching AI Watching You

Every Single Day, And Every Word You Say

My Most Recent Podcast

My most recent guest on Brave New World was Arthur Spirling, Professor of Political Science at Princeton University. Arthur’s recent work focuses on understanding how Artificial Intelligence is changing the relationship between governments and their citizens. 

As I have written previously, governments around the world have adopted very different models of Internet use, ranging from authoritarian to market driven. Who controls AI is a big looming issue confronting regulators.

Arthur and I talked about a range of topics, such as how measurement has changed in political science with the availability of data, and its implications for governance and accountability.   For example, democratic governments must serve their citizens by releasing previously censored data to the public periodically as a way of maintaining transparency and trust between the government and its citizens.

So, check out my conversation with Arthur for more color on AI and politics:

https://bravenewpodcast.com/episodes/2024/03/07/episode-79-arthur-spirling-on-how-ai-can-change-politics/

AI In Government

Arthur made a parenthetical remark during our conversation that stayed with me, namely, that the US government censors more data than it should, and is very slow about releasing such data to the public. For example, police records are increasingly hard to obtain, and the government is even dragging its feet on declassifying diplomatic communications.

Why is the government withholding more information even as it collects more of it? Will AI worsen this divide?

While there isn’t sufficient data to answer this question definitively, I think it will. Precedence suggests that government officials will be tempted to use the power of AI, but will cover their tracks. The US provides some compelling examples of such behavior from the Reagan administration, when computers were just making their way into government.

Computers transformed paper-based policing and criminal justice in the 1980s.  Former Deputy Administrator Charles Work of the Department of Justice (DOJ) remarked how often a prosecutor would walk down a hall, bump into a colleague, and discover by accident that there was another case ongoing against the same defendant. Work introduced the first Prosecutors Management Information System (PROMIS) to track cases, defendants, and charges. PROMIS was developed by a company named Inslaw, whose founder, Bill Hamilton, had worked extensively with US Intelligence agencies prior to his work with the Justice department.

PROMIS changed everything. Prosecutors now had unprecedented ability to trace and link cases, crimes and people. Computers brought transparency and efficiency. Information became liquid, in the sense that it became easier to obtain and link with other sources to get insights that were previously impossible.

And then suddenly, the Justice department went rogue. It forced Inslaw to hand over the source code and drove the business into bankruptcy by withholding payments. By all accounts, it stole the software, modified it, allegedly by installing a back-door to surveil its use.

Inslaw sued the DOJ.

Federal judge George Bason, who presided over the Inslaw-DOJ case issued a ruling in 1987 ordering the DOJ to pay Inslaw 6.8 million, noting that “the DOJ took, converted, stole, INSLAW's enhanced PROMIS by trickery, fraud, and deceit."  Bason found that DOJ had acted in bad faith, and called the testimony of DOJ witnesses biased, unbelievable, and unreliable. The DOJ officials displayed “collective amnesia,” he said.

In an extraordinary development, Bason was then removed from the bench and replaced by the DOJ attorney who had lost the case. The source code of the modified PROMIS system was awarded to an outfit associated with Earl Brian, a longtime aide to Ronald Reagan from his days as the Governor of California. Brian allegedly sold it to financial institutions and foreign governments. A decade later, in October 1996, a federal jury in Los Angeles convicted Brian on 10 counts of criminal behavior, including conspiracy and securities fraud. He was sentenced to a four-and-a-half-year prison term. The jury found Brian to be a con man. What were he and the Reagan administration up to?

Who Does AI Work For?

A recent documentary called “American Conspiracy: The Octopus Files,” traces the DOJ’s history with PROMIS through the eyes of investigative journalist Danny Casalaro in the early 80s. The four-episode series begins with Inslaw and the PROMIS system, and follows Casalaro’s investigative journey that threatened to expose the corruption and scandals that rocked the Reagan administration.

Casalaro was found dead in his hotel room bathtub in Martinsburg, West Virginia in August 1991 after he had spent a decade uncovering the tentacles that spread far and wide from the PROMIS case. He had gone to Martinsburg to meet one last potential source for his book. According to the notes he left behind in an archive at the University of Missouri, Casalaro had evidently connected the dots between PROMIS, the DOJ, the Reagan administration, clandestine agencies, and organized crime during the 1980s.

The death was immediately ruled a suicide and no autopsy was performed despite various red flags. Casalaro had suffered eight cuts in one arm and four in the other arm deep enough to sever his tendons.  Seasoned paramedic Don Shirley who pulled Casalaro out the tub had this to say about the suicide theory: “he could not have physically done that. The cuts were so deep that the tendons had been severed. You can’t ignore facts. If you cut your tendons, you can’t hold somethin’. Those are the simple facts.”

Watch the documentary. The larger issue it raised for me is that in the age of AI machines with unfettered access to information, regulating the regulator is arguably one of the most vexing problems facing us. An unregulated regulator can and will go rogue and cover its tracks from public scrutiny. This lack of transparency erodes trust in democratic institutions and in democracy itself. In their book titled The Narrow Corridor,  political economists Daren Acemoglu and James Robinson present historical data arguing that a democracy becomes stronger by shackling the power of its government, where the shackles serve as a mechanism against misuse. This bolsters public confidence in the state.

One of the ongoing concerns about AI is that it could go rogue on us if its goals become misaligned with ours. While this is a legitimate concern, the larger threat at the moment is from people who control the AI going rogue. There’s no one watching them.