Connect with us

Artificial Intelligence

Criminal Justice Reform Advocates Agree that Current AI Assessment Tools are Garbage, But Differ on How to Proceed

Published

on

Photo of Brookings Institution panelists by David Jelke

WASHINGTON, February 20, 2020— Measures designed to assess the likelihood of an individual’s recidivism, and to consequently decide whether a defendant awaiting trial can be released before trial, are ineffective.

That was the conclusion of panelists at a Brookings Institution event titled “AI, Predictive Analytics, and Criminal Justice.” They generally agreed that “we have really poorly designed measures” of assessment.

All panelists took turns dumping on new technologies designed by companies hired by the government to assess recidivism probabilities.

These algorithms took into account personal factors such as number of prior arrests and severity.

Sakira Cook, director of the Justice Reform Program at The Leadership Conference on Civil and Human Rights, roundly attacked the fairness of these tools in three ways, disputing the claim that they aid law enforcement in making decisions regarding bail. First, she outlined the distinction between “crime data,” which is the term that is often used by industry know-hows, and “arrest data,” which is the more accurate term, since the dataset includes arrests by law enforcement that don’t lead to conviction.

Secondly, Cook criticized the lack of transparency surrounding these tools. “Certain government agreements” with licensors prevent researchers or activists from accessing the proprietary data without signing a Non-Disclosure Agreement. “These tools are not transparent,” contended Cook.

Thirdly, and more broadly, Cook belittled AI tools by attacking the foundation they rest on— the supposed legitimacy of U.S. criminal law. “We agree that we have to change the fundamental laws of this country,” said Cook, referencing the disproportionate amount of black men who are arrested and incarcerated in the U.S.

Faye Taxman, Ph.D., professor at George Mason University, agreed with the notion of a major judicial review. However, she made a point to defend the idea of these AI-based tools.

The tools’ biggest problems, Taxman asserted, are the variables being fed into the algorithms.

She gave the hypothetical example of a 30-year-old who is identified by the technology as having a “drug problem” for smoking marijuana in high school. These “lifetime variables” are one of several foul data points that need to be edited out of future generations of tools.

Taxman also emphasized the importance of tools over nothing, referencing how the first generation of “tools” were just prison psychologists in the 1920s who decided the likelihood of inmate recidivism based on personal hunches and flimsy science.

“Do we need instruments at all? As a scientist, I say we need instruments,” said Taxman.

David Jelke was a Reporter for Broadband Breakfast. He graduated from Dartmouth College with a degree in neuroscience. Growing up in Miami, he learned to speak Spanish during a study abroad semester in Peru. He is now teaching himself French on his iPhone.

Artificial Intelligence

AI Should Compliment and Not Replace Humans, Says Stanford Expert

AI that strictly imitates human behavior can make workers superfluous and concentrate power in the hands of employers.

Published

on

Photo of Erik Brynjolfsson, director of the Stanford Digital Economy Lab, in January 2017 by Sandra Blaser used with permission

WASHINGTON, November 4, 2022 – Artificial intelligence should be developed primarily to augment the performance of, not replace, humans, said Erik Brynjolfsson, director of the Stanford Digital Economy Lab, at a Wednesday web event hosted by the Brookings Institution.

AI that complements human efforts can increase wages by driving up worker productivity, Brynjolfsson argued. AI that strictly imitates human behavior, he said, can make workers superfluous – thereby lowering the demand for workers and concentrating economic and political power in the hands of employers – in this case the owners of the AI.

“Complementarity (AI) implies that people remain indispensable for value creation and retain bargaining power in labor markets and in political decision-making,” he wrote in an essay earlier this year.

What’s more, designing AI to mimic existing human behaviors limits innovation, Brynjolfsson argued Wednesday.

“If you are simply taking what’s already being done and using a machine to replace what the human’s doing, that puts an upper bound on how good you can get,” he said. “The bigger value comes from creating an entirely new thing that never existed before.”

Brynjolfsson argued that AI should be crafted to reflect desired societal outcomes. “The tools we have now are more powerful than any we had before, which almost by definition means we have more power to change the world, to shape the world in different ways,” he said.

The AI Bill of Rights

In October, the White House released a blueprint for an “AI Bill of Rights.” The document condemned algorithmic discrimination on the basis of race, sex, religion, or age and emphasized the importance of user privacy. It also endorsed system transparency with users and suggested the use of human alternatives to AI when feasible.

To fully align with the blueprint’s standards, Russell Wald, policy director for Stanford’s Institute for Human-Centered Artificial Intelligence, argued at a recent Brookings event that the nation must develop a larger AI workforce.

Continue Reading

Artificial Intelligence

Workforce Training Needed to Address Artificial Intelligence Bias, Researchers Suggest

Building on the Blueprint for an AI Bill of Rights by the White House Office of Science and Technology Policy.

Published

on

Russell Wald. Credit: Rod Searcey, Stanford Law School

WASHINGTON, October 24, 2022–To align with the newly released White House guide on artificial intelligence, Stanford University’s director of policy said at an October Brookings Institution event last week that there needs to be more social and technical workforce training to address artificial intelligence biases.

Released on October 4, the Blueprint for an AI Bill of Rights framework by the White House’s Office of Science and Technology Policy is a guide for companies to follow five principles to ensure the protection of consumer rights from automated harm.

AI algorithms rely on learning the users behavior and disclosed information to customize services and advertising. Due to the nature of this process, algorithms carry the potential to send targeted information or enforce discriminatory eligibility practices based on race or class status, according to critics.

Risk mitigation, which prevents algorithm-based discrimination in AI technology is listed as an ‘expectation of an automated system’ under the “safe and effective systems” section of the White House framework.

Experts at the Brookings virtual event believe that workforce development is the starting point for professionals to learn how to identify risk and obtain the capacity to fulfill this need.

“We don’t have the talent available to do this type of investigative work,” Russell Wald, policy director for Stanford’s Institute for Human-Centered Artificial Intelligence, said at the event.

“We just don’t have a trained workforce ready and so what we really need to do is. I think we should invest in the next generation now and start giving people tools and access and the ability to learn how to do this type of work.”

Nicol Turner-Lee, senior fellow at the Brookings Institution, agreed with Wald, recommending sociologists, philosophers and technologists get involved in the process of AI programming to align with algorithmic discrimination protections – another core principle of the framework.

Core principles and protections suggested in this framework would require lawmakers to create new policies or include them in current safety requirements or civil rights laws. Each principle includes three sections on principles, automated systems and practice by government entities.

In July, Adam Thierer, senior research fellow at the Mercatus Center of George Mason University stated that he is “a little skeptical that we should create a regulatory AI structure,” and instead proposed educating workers on how to set best practices for risk management, calling it an “educational institution approach.”

Continue Reading

Artificial Intelligence

Deepfakes Pose National Security Threat, Private Sector Tackles Issue

Content manipulation can include misinformation from authoritarian governments.

Published

on

Photo of Dana Roa of Adobe, Paul Lekas of Global Policy (left to right)

WASHINGTON, July 20, 2022 – Content manipulation techniques known as deepfakes are concerning policy makers and forcing the public and private sectors to work together to tackle the problem, a Center for Democracy and Technology event heard on Wednesday.

A deepfake is a technical method of generating synthetic media in which a person’s likeness is inserted into a photograph or video in such a way that creates the illusion that they were actually there. Policymakers are concerned that deepfakes could pose a threat to the country’s national security as the technology is being increasingly offered to the general population.

Deepfake concerns that policymakers have identified, said participants at Wednesday’s event, include misinformation from authoritarian governments, faked compromising and abusive images, and illegal profiting from faked celebrity content.

“We should not and cannot have our guard down in the cyberspace,” said Representative John Katko, R-NY, ranking member of House Committee on homeland security.

Adobe pitches technology to identify deepfakes

Software company Adobe released an open-source toolkit to counter deepfake concerns earlier this month, said Dana Rao, executive vice president of Adobe. The companies’ Content Credentials feature is a technology developed over three years that tracks changes made to images, videos, and audio recordings.

Content Credentials is now an opt-in feature in the company’s photo editing software Photoshop that it says will help establish credibility for creators by adding “robust, tamper-evident provenance data about how a piece of content was produced, edited, and published,” read the announcement.

Adobe’s Connect Authenticity Initiative project is dedicated to addressing problems establishing trust after the damage caused by deepfakes. “Once we stop believing in true things, I don’t know how we are going to be able to function in society,” said Rao. “We have to believe in something.”

As part of its initiative, Adobe is working with the public sector in supporting the Deepfake Task Force Act, which was introduced in August of 2021. If adopted, the bill would establish a National Deepfake and Digital task force comprised of members from the private sector, public sector, and academia to address disinformation.

For now, said Cailin Crockett, senior advisor to the White House Gender Policy Council, it is important to educate the public on the threat of disinformation.

Continue Reading

Signup for Broadband Breakfast

Get twice-weekly Breakfast Media news alerts.
* = required field

Broadband Breakfast Research Partner

Trending