UMass Amherst and Stanford team designs new framework for fairer, safer algorithms
Seventy years ago, science fiction writer Isaac Asimov imagined a world where robots would serve humans in countless ways, and he equipped them with built-in safeguards now known as Asimov’s Three Laws of Robotics, to prevent them, among other goals, from ever harming a person.
Guaranteeing safe and fair machine behavior is still an issue today, says machine learning researcher and lead author Philip Thomas at the University of Massachusetts Amherst. “When someone applies a machine learning algorithm, it’s hard to control its behavior,” he points out. This risks undesirable outcomes from algorithms that direct everything from self-driving vehicles to insulin pumps to criminal sentencing, say he and co-authors.
Writing in Science, Thomas and his colleagues Yuriy Brun, Andrew Barto and graduate student Stephen Giguere at UMass Amherst, Bruno Castro da Silva at the Federal University of Rio Grande del Sol, Brazil, and Emma Brunskill at Stanford University this week introduce a new framework for designing machine learning algorithms that make it easier for users of the algorithm to specify safety and fairness constraints.
“We call algorithms created with our new framework ‘Seldonian’ after Asimov’s character Hari Seldon,” Thomas explains. “If I use a Seldonian algorithm for diabetes treatment, I can specify that undesirable behavior means dangerously low blood sugar, or hypoglycemia. I can say to the machine, ‘while you’re trying to improve the controller in the insulin pump, don’t make changes that would increase the frequency of hypoglycemia.’ Most algorithms don’t give you a way to put this type of constraint on behavior; it wasn’t included in early designs.”
“But making it easier to ensure fairness and avoid harm is becoming increasingly important as machine learning algorithms impact our lives more and more,” he says.
However, “a recent paper listed 21 different definitions of fairness in machine learning. It’s important that we allow the user to select the definition that is appropriate for their intended application,” he adds. “The interface that comes with a Seldonian algorithm allows the user to do just this: to define what ‘undesirable behavior’ means for their application.”
In Asimov’s Foundation series, Seldon is in the same universe as his Robot series. Thomas explains, “Everything has fallen apart, the galactic empire is collapsing, partly because the Three Laws of Robotics require certainty. With that level of safety required, robots are paralyzed with indecision because they cannot act with certainty and guarantee that no human will be harmed by their actions.”
Seldon proposes fixing this by turning to reasoning probabilistically about safety. “That’s a good fit to what we’re doing, Thomas says. The new approach he and colleagues provide allows for probabilistic constraints and requires the algorithm to specify ways the user can tell it what to constrain. He says, “The framework is a tool for the machine learning researcher. It guides them toward creating algorithms that are easier for users to apply responsibly to real-world problems.”
To test the new framework, they applied it to predict grade point averages in a data set of 43,000 students in Brazil by creating a Seldonian algorithm with constraints. It successfully avoided several types of undesirable gender bias. In another test, they show how an algorithm could improve the controller in an insulin pump while guaranteeing that it would not increase the frequency of hypoglycemia.
Thomas says, “We believe there’s massive room for improvement in this area. Even with our algorithms made of simple components, we obtained impressive results. We hope that machine learning researchers will go on to develop new and more sophisticated algorithms using our framework, which can be used responsibly for applications where machine learning used to be considered too risky. It’s a call to other researchers to conduct research in this space.”
Learn more: New Machine Learning Algorithms Offer Safety and Fairness Guarantees
The Latest Google Headlines on:
Artificial intelligence algorithms
[google_news title=”” keyword=”Artificial intelligence algorithms” num_posts=”10″ blurb_length=”0″ show_thumb=”left”]
The Latest Bing News on:
Artificial intelligence algorithms
- Artificial intelligence reshaping HR practiceson May 8, 2024 at 3:01 pm
Across the globe and in today’s rapidly evolving technological landscape, the integration of artificial intelligence (AI) is reshaping traditional human resource (HR) practices.
- Why are algorithms called algorithms? A brief history of the Persian polymath you've likely never heard ofon May 8, 2024 at 8:42 am
Algorithms have become integral to our lives. From social media apps to Netflix, algorithms learn your preferences and prioritize the content you are shown. Google Maps and artificial intelligence are ...
- Artificial intelligence in banking has strong adoption by “data-first” FIson May 8, 2024 at 12:30 am
In the rapidly evolving world of digital banking, the financial services industry stands at the cusp of a revolution, powered by Artificial Intelligence ...
- Quantes: Leading British Crypto Fund, Uniting Artificial Intelligence and Tradingon May 7, 2024 at 12:47 am
In the modern world of finance and technology, new players are emerging, ready to change the landscape of investments. One such player is Quantes, a British crypto fund that has been successfully ...
- How artificial intelligence is helping Ghana plan for a renewable energy futureon May 6, 2024 at 4:59 pm
The technology is helping the West African nation to invest wisely in infrastructure, prioritising energy and food security, but also human health.
- Artificial Intelligence, Psychedelics, and Psychotherapy Working Together to Fight Chronic Painon May 6, 2024 at 4:45 pm
Western medicine has long treated chronic pain by focusing only on pain receptors; two Dallas scientists are working to rewrite this approach.
- 1 Artificial Intelligence (AI) Stock That Could Be Worth $2 Trillion By 2030on May 6, 2024 at 7:31 am
The stock recently dropped sharply following the release of its first-quarter earnings report, but that shouldn't scare off investors.
- A physicists’ guide to the ethics of artificial intelligenceon May 6, 2024 at 6:00 am
Physics may seem like its own world, but different sectors using machine learning are all part of the same universe.
- Using Artificial Intelligence To Think Outside The Boxon May 3, 2024 at 3:05 pm
The enablement comes in the form of “rapidly analyzing the innovation pipeline to discover windows of opportunities," McKnight continues. "Algorithms can collect and connect relevant data on ...
- Artificial Intelligence Algorithms Are Not Clairvoyanton May 2, 2024 at 8:19 am
Artificial intelligence (AI) systems, and computers in general, possess several advantages over humans. They have virtually perfect recall and are not subject to fatigue, mood variations, or ...
The Latest Google Headlines on:
Safe and fair machine behavior
[google_news title=”” keyword=”Safe and fair machine behavior” num_posts=”10″ blurb_length=”0″ show_thumb=”left”]
The Latest Bing News on:
Safe and fair machine behavior
- Klaviyo (KVYO) Q1 2024 Earnings Call Transcripton May 8, 2024 at 6:45 pm
Good afternoon, and welcome to Klaviyo's first quarter of fiscal 2024 earnings conference call. [Operator instructions] With that, I'd like to turn the call over to Jack Grant, senior director of ...
- Dutch Online Gambling: What the Future Holds?on May 8, 2024 at 11:47 am
The iGaming industry enjoys one of the most successful markets in 2024, especially online. It seems that gambling fans have become a lot more interested in online casinos, and for good reason.
- Interview: Suzanne Gildert leaves Sanctuary to focus on AI consciousnesson April 30, 2024 at 2:26 am
Earlier this month, co-founder and CTO Suzanne Gildert left humanoid robotics company Sanctuary AI and her beloved Phoenix robot to focus full-time on AI safety, ethics and consciousness. We spoke to ...
- Trump hush money trial closer to opening statements after 12 jurors selectedon April 18, 2024 at 3:56 pm
The first juror who said that she was concerned essentially about her privacy and I think implicitly her safety because she ... suggests they might not be a fair and impartial juror.
- Who Really Owns AI Art? What Businesses Need To Know About Generative AI And Intellectual Ownershipon April 16, 2024 at 6:00 am
The argument is that centuries of societal prejudices and behavior will inevitably show ... in every arena where generative AI exists. It’s safe to say the art world, much like the written ...
- In Australia, ‘Cats Are Just Catastrophic’on April 15, 2024 at 5:00 pm
The automated, solar-powered machine ... reserves and other safe havens may exacerbate the problem, by making it safe for sheltered populations to lose whatever defensive behaviors they did ...
- Instagram to crack down on teen sextortionon April 11, 2024 at 1:00 am
The company told CNN that Meta’s technology uses on-device machine learning to ... information about accounts and behaviors that violate their child safety. The company said it is now adding ...
- The 15 Best Dog Collars of 2024, Tested and Reviewedon January 10, 2024 at 12:52 pm
The Orvis Reflective Dog Collar is comfortable, safe, and built to last It’s easy to get caught ... The polyester webbing proved easy to clean (our tester threw it in the washing machine) and the ...
- The FTC Intervenes In The AI Safety Race Between OpenAI And Anthropicon July 14, 2023 at 8:28 am
Kahn did promise to enforce fair AI competition and ... documents like a constitution. My behaviors and responses are determined by my training data and machine learning model, not any set of ...
- Law and Crimeon May 3, 2022 at 9:47 am
In the courtroom, legal practitioners require a grasp of defendants' motivations and actions in order to render fair judgment ... psychologists study the behaviors and motivations of criminals.