Policymakers often draw on the work of social scientists to predict how specific policies might affect social outcomes such as the employment or crime rates. The idea is that if they can understand how different factors might change the trajectory of someones life, they can propose interventions to promote the best outcomes.
In recent years, though, they have increasingly relied upon machine learning, which promises to produce far more precise predictions by crunching far greater amounts of data. Such models are now used to predict the likelihood that a defendant might be arrested for a second crime, or that a kid is at risk for abuse and neglect at home. The assumption is that an algorithm fed with enough data about a given situation will make more accurate predictions than a human or a more basic statistical analysis.
Sign up for The Algorithm artificial intelligence, demystified
Now a new study published in the Proceedings of the National Academy of Sciences casts doubt on how effective this approach really is. Three sociologists at Princeton University asked hundreds of researchers to predict six life outcomes for children, parents, and households using nearly 13,000 data points on over 4,000 families. None of the researchers got even close to a reasonable level of accuracy, regardless of whether they used simple statistics or cutting-edge machine learning.
The study really highlights this idea that at the end of the day, machine-learning tools are not magic, says Alice Xiang, the head of fairness and accountability research at the nonprofit Partnership on AI.
The researchers used data from a 15-year-long sociology study called the Fragile Families and Child Wellbeing Study, led by Sara McLanahan, a professor of sociology and public affairs at Princeton and one of the lead authors of the new paper. The original study sought to understand how the lives of children born to unmarried parents might turn out over time. Families were randomly selected from children born in hospitals in large US cities during the year 2000. They were followed up for data collection when the children were 1, 3, 5, 9, and 15 years old.
McLanahan and her colleagues Matthew Salganik and Ian Lundberg then designed a challenge to crowdsource predictions on six outcomes in the final phase that they deemed sociologically important. These included the childrens grade point average at school; their level of grit, or self-reported perseverance in school; and the overall level of poverty in their household. Challenge participants from various universities were given only part of the data to train their algorithms, while the organizers held some back for final evaluations. Over the course of five months, hundreds of researchers, including computer scientists, statisticians, and computational sociologists, then submitted their best techniques for prediction.
The fact that no submission was able to achieve high accuracy on any of the outcomes confirmed that the results werent a fluke. You can't explain it away based on the failure of any particular researcher or of any particular machine-learning or AI techniques, says Salganik, a professor of sociology. The most complicated machine-learning techniques also werent much more accurate than far simpler methods.
For experts who study the use of AI in society, the results are not all that surprising. Even the most accurate risk assessment algorithms in the criminal justice system, for example, max out at 60% or 70%, says Xiang. Maybe in the abstract that sounds somewhat good, she adds, but reoffending rates can be lower than 40% anyway. That means predicting no reoffenses will already get you an accuracy rate of more than 60%.
Likewise, research has repeatedly shown that within contexts where an algorithm is assessing risk or choosing where to direct resources, simple, explainable algorithms often have close to the same prediction power as black-box techniques like deep learning. The added benefit of the black-box techniques, then, is not worth the big costs in interpretability.
The results do not necessarily mean that predictive algorithms, whether based on machine learning or not, will never be useful tools in the policy world. Some researchers point out, for example, that data collected for the purposes of sociology research is different from the data typically analyzed in policymaking.
Rashida Richardson, policy director at the AI Now institute, which studies the social impact of AI, also notes concerns in the way the prediction problem was framed. Whether a child has grit, for example, is an inherently subjective judgment that research has shown to be a racist construct for measuring success and performance, she says. The detail immediately tipped her off to thinking, Oh theres no way this is going to work.
Salganik also acknowledges the limitations of the study. But he emphasizes that it shows why policymakers should be more careful about evaluating the accuracy of algorithmic tools in a transparent way. Having a large amount of data and having complicated machine learning does not guarantee accurate prediction, he adds. Policymakers who don't have as much experience working with machine learning may have unrealistic expectations about that.
To have more stories like this delivered directly to your inbox,sign upfor our Webby-nominated AI newsletter The Algorithm. It's free.
Read the original post:
- What Is Artificial Intelligence (AI)? | PCMag - July 2nd, 2020
- How AI Is Impacting Operations At LinkedIn - Forbes - July 2nd, 2020
- Does the Human Touch + AI = The Future of Work? - Datanami - July 2nd, 2020
- E-Commerce and AI Working Together in a Post COVID19 World - IoT For All - July 2nd, 2020
- AI Models Say These Are The Best Stocks To Short Today - Forbes - July 2nd, 2020
- Data Quantity, Complexity Drives Use of AI in Drug Discovery and Testing - Xconomy - July 2nd, 2020
- Council of Europe participating in the panel on Accountable AI solutions of the AI for Good Global Summit 2020 - Council of Europe - July 2nd, 2020
- The DOD needs to define AI and protect its data better, watchdog says - FedScoop - July 2nd, 2020
- 3 AI Healthcare Stocks to Ride Out the Coronavirus Resurgence - Yahoo Finance - July 2nd, 2020
- MIT removes huge dataset that teaches AI systems to use racist, misogynistic slurs - The Next Web - July 2nd, 2020
- Remdesivir's controversial cost, early vaccine data, and AI at the end of life - STAT - July 2nd, 2020
- Artificial intelligence levels show AI is not created equal. Do you know what the vendor is selling? - Spend Matters - July 2nd, 2020
- China and AI: What the World Can Learn and What It Should Be Wary of - Nextgov - July 2nd, 2020
- Machines That Can Understand Human Speech: The Conversational Pattern Of AI - Forbes - July 2nd, 2020
- How Can Companies Overcome Challenges That Come With AI Adoption? - Express Computer - July 2nd, 2020
- Google Sheet Comes Up With A New Feature AI-Backed Smart Fill and Smart Clean Up That Will Make The Data Error Free And Consistent - Digital... - July 2nd, 2020
- Artificial Intelligence Systems Will Need to Have Certification, CISA Official Says - Nextgov - July 2nd, 2020
- Artificial Intelligence Can't Deal With Chaos, But Teaching It Physics Could Help - ScienceAlert - July 2nd, 2020
- NASAs New Moon-Bound Space Suits Will Get a Boost From AI - WIRED - July 2nd, 2020
- China & AI: What the World Can Learn And What It Should Be Wary of - The Quint - July 2nd, 2020
- Artificial Intelligence (AI) in Agriculture Market: Technological Innovations and Analysis Till 2030 - Cole of Duty - July 2nd, 2020
- Letters to the editor - The Economist - July 2nd, 2020
- DHL partners with Microsoft and AI specialist to ease automation integration - Parcel and Postal Technology International - July 2nd, 2020
- Coronavirus will finally give artificial intelligence its moment - San Antonio Express-News - June 20th, 2020
- AI Makes A Splash In Advertising - Forbes - June 20th, 2020
- Artificial Intelligence in Fintech - Global Market Growth, Trends and Forecasts to 2025 - Assessment of the Impact of COVID-19 on the Industry -... - June 20th, 2020
- This Startup Is Trying to Foster an AI Art Scene in Korea - Adweek - June 20th, 2020
- 3 examples that show how automation and AI are not just for big businesses - AZ Big Media - June 20th, 2020
- AI cameras may be used to detect social distancing as US is reopening - Business Insider - Business Insider - June 20th, 2020
- NIST Announces Workshop Focused on "Bias in AI" - JD Supra - June 20th, 2020
- The importance and opportunities of transatlantic cooperation on AI - Brookings Institution - June 20th, 2020
- AI bias and privacy issues require more than clever tech - ComputerWeekly.com - June 20th, 2020
- Is FaceApp Just a Gender Swapping Gimmick or an AI and Privacy Nightmare in the Making? - Wccftech - June 20th, 2020
- Where AI Meets Big Data, Innovation is Assured - Analytics Insight - June 20th, 2020
- AI Can Autocomplete Images in the Same Way It Does Text - Adweek - June 20th, 2020
- Amazons AI-powered distance assistants will warn workers when they get too close - The Verge - June 20th, 2020
- OODA Video: The Technologies of AI Security and Ethics - OODA Loop - June 20th, 2020
- The UK has only just begun to see the transformative potential of AI - Telegraph.co.uk - June 20th, 2020
- US Corporations are talking about bans for AI. Will the EU? - EURACTIV - June 20th, 2020
- 'Lost memories': War crimes evidence threatened by AI moderation - Reuters - June 20th, 2020
- Tesla's Head of AI Says The Firm Uses a Harder Approach to Self-Driving for Scalability Reasons - Interesting Engineering - June 20th, 2020
- AI in Healthcare Market projected a CAGR of 52.3% during the forecast period, 2020-2026 - 3rd Watch News - June 20th, 2020
- 'Its fundamental': Graphcore CEO believes new kinds of AI will prove the worth of a new kind of computer - ZDNet - June 1st, 2020
- Microsoft lays off journalists to replace them with AI - The Verge - June 1st, 2020
- Can't find Wally? AI to the rescue... - AI Daily - June 1st, 2020
- AI is more data-hungry than ever, and DefinedCrowd raises $50M B round to feed it - TechCrunch - June 1st, 2020
- Anoto's subsidiary Knowledge AI Inc expands the customer base for its education solution KAIT by signing a fully paid subscription agreement with a... - June 1st, 2020
- Coronavirus tests the value of artificial intelligence in medicine - FierceHealthcare - June 1st, 2020
- Walmart Employees Are Out to Show Its Anti-Shoplifting AI Doesn't Work - WIRED - June 1st, 2020
- Brabant universities designed an AI-Toolkit to support the industry - Innovation Origins - June 1st, 2020
- Experts discuss benefits and risks associated with integrating AI - The Peninsula Qatar - June 1st, 2020
- (2020-2025) AI in Social Media Market Insights and Industry Detailed Analysis: Google, Facebook, Microsoft, AWS, IBM, Adobe Systems, Baidu,... - June 1st, 2020
- AI and the Far Right: A History We Can't Ignore - Medium - May 4th, 2020
- Jukebox AI trained to create entires songs that are actually good - Business Insider - Business Insider - May 4th, 2020
- A critical review of Star Wars AI - The Next Web - May 4th, 2020
- How Should Your Business Get Started With AI? Begin With Why - Forbes - May 4th, 2020
- Machine AI Is On The Race to Overtake Human AI - Analytics Insight - May 4th, 2020
- The impact of artificial intelligence on intelligence analysis - Reuters - May 4th, 2020
- The Impending Artificial Intelligence Revolution in Healthcare - Op-Ed - HIT Consultant - May 4th, 2020
- Discover how AI can help with early detection of glaucoma - Health Europa - May 4th, 2020
- OSS to Host AI at the Edge Webinar with Leaders from NVIDIA and Marvell - AiThority - May 4th, 2020
- How AI is changing the customer experience - MIT Technology Review - May 4th, 2020
- How one Microsoft mom inspired health care companies to embrace the life-saving potential of AI | Transform - Microsoft - May 4th, 2020
- A.I. can't solve this: The coronavirus could be highlighting just how overhyped the industry is - CNBC - May 4th, 2020
- Determined AI makes its machine learning infrastructure free and open source - TechCrunch - May 4th, 2020
- Adidas turns to AI and 3D printing to boost sales during coronavirus - Business Insider - Business Insider - May 4th, 2020
- This startup by San Francisco-based founders uses AI to find the right hires for corporates - YourStory - May 4th, 2020
- Neural net-generated memes are one of the best uses of AI on the internet - The Verge - May 4th, 2020
- Rehearsal Announces Major Release Incorporating Self-Guided Practice and Artificial Intelligence - PR Web - May 4th, 2020
- How AI Can Help Pharma Risk Managers Beat Their Regulatory Nightmares and Get the Rest They Need - Workers Comp Forum - May 4th, 2020
- Doctors are using AI to triage covid-19 patients. The tools may be here to stay - MIT Technology Review - April 26th, 2020
- Open Resources to Become Knowledgeable in the Field of AI - ArchDaily - April 26th, 2020
- Reducing the carbon footprint of artificial intelligence - MIT News - April 26th, 2020
- AI used to predict Covid-19 patients' decline before proven to work - STAT - April 26th, 2020
- AI could transform open source intelligence in the developing world - C4ISRNet - April 26th, 2020
- This AI wrote such emo lyrics that humans thought it was My Chemical Romance - The Next Web - April 26th, 2020
- The buzzword is AI - The Hindu - April 26th, 2020
- AI Weekly: AI models illustrate the importance of continued social distancing - VentureBeat - April 26th, 2020
- Someone taught an AI to draw dicks after feeding it 25,000 doodles of penises - PC Gamer - April 26th, 2020
- Facebook, AWS team up to produce open-source PyTorch AI libraries, grad student says he successfully used GPT-2 to write his homework.... - The... - April 26th, 2020