Welcome back. After spending billions of dollars on development, IBM Corp. is looking to offload Watson Health, its costly and unprofitable AI-powered healthcare unit. The troubled business, which has helped create tools to analyze mammograms, improve patient communications and parse complex healthcare data, highlights broader challenges in the health-tech market. A key barrier for AI developers is the lack of access to patient data, stemming from privacy issues. By contrast, in sectors where data is plentiful and readily accessible, such as financial markets or logistics, advances continue to accelerate.

Amid a global healthcare crisis, International Business Machines Corp. is exploring the sale of Watson Health, an unprofitable business unit created six years ago to develop AI-enabled tools for doctors and other medical practitioners, The Wall Street Journal reports. 

Bad medicine. Watson Health was one of IBM’s first and largest bets on AI, an area Chief Executive Arvind Krishna says is pivotal for the company’s success, along with its cloud-computing business.

Data privacy barriers. Like Watson Health, Alphabet Inc.’s Google DeepMind unit has also faced challenges navigating patient privacy issues with its own healthcare initiatives, which have lost money in recent years.

Despite its troubles in healthcare, IBM has pushed forward in other areas, including striking a multiyear contract to help Delta Air Lines Inc. shift its data and applications into the cloud, in part to access cloud-based artificial intelligence tools, WSJ’s Jared Council reports. 

Learning to fly. Under the deal, IBM will train more than 1,000 Delta IT workers in how to operate in cloud environments, including helping them develop skills related to application development, data management and security.

Bank of New York Mellon Corp., working with Google Cloud, has developed machine-learning technology that is designed to predict when Treasury market trades will fail to settle, due to a lack of securities or other issues, WSJ's John McCormick reports.

Failing trades. After testing a software model for 10 months using settlement data, the bank found it can spot about 40% of trades that are likely to fail.

Settling accounts. BNY Mellon, which is one of the largest clearing and settlement firms in the roughly $20 trillion Treasury market, said that on a typical day about 2% of transactions fail to settle.

J.B. Hunt Transport Services Inc. is turning to Google’s AI to better pair shippers with carriers, aiming to help shippers see supply-and-demand needs days ahead and carriers to estimate future fuel and other transportation costs, John McCormick reports. 

Digital competitors. Spurred by tech startups and digitally-driven market entrants like Uber Technologies Inc., traditional trucking and logistics players are investing heavily in technology to keep pace with rapid changes in the market and not fall behind.

Join us on March 31 for the WSJ Pro AI Forum, with panel discussions and interviews with corporate thought leaders, case studies, breakout groups and more. Topics will include reinventing the supply chain, AI in medicine and the revolution in tech regulation. More information is available here. 

Big data-center deal. Reflecting the rapid growth of the data-center market, driven in part by increasing demand for computing power for data-heavy smart software, Starboard Value Acquisition Corp. is in talks to buy data-center provider Cyxtera Technologies Inc. for $3.4 billion, people familiar with the deal say. (The Wall Street Journal) 

First church of AI closes. Anthony Levandowski, a former Google engineer who was pardoned by President Trump after being convicted of stealing self-driving car secrets, is closing the Way of the Future church, which he launched in 2015 as a religious sect that worships AI. (TechCrunch) 

No small matter. Tampere University researchers used AI to predict what happens when ultrashort light pulses interact with matter, a capability that can help optimize lasers used in surgery or manufacturing automation. (TechXplore) 

Real researcher, fake co-author. David Cox, the co-director at a high-profile Cambridge, Mass., AI lab, says his name and photo were included in a pair of Chinese computer-science research papers that credited him as a co-author, even though he had nothing to do with them. (Wired) 

Google fires another AI watchdog. Two months after firing a high-profile AI researcher, Google confirmed this weekend that it had fired Margaret Mitchell, a second AI ethics leader, saying the move came “after conducting a review of this manager’s conduct.” (The New York Times)

Comparing apples and oranges. Israeli startup Tevel Aerobotics Technologies has developed an autonomous drone that can differentiate and pick ripe fruit, a capability pursued in direct response to labor shortages during the pandemic, the company said. (The World Economic Forum) 

Jobs are safe, busywork at risk. Steve Shwartz, an AI researcher and investor, argues in a new book that the notion of AI taking jobs is a myth, saying smart systems are only capable of learning tasks that involve repetitive data-oriented functions. (Forbes)

Better bedside manner. Tech startup Virti is using AI to create virtual patients designed to help doctors and other hospital staff improve their communications and interpersonal skills with patients, via software that can be downloaded onto a smartphone or laptop. (The Washington Post) 

Software for hard bodies. A Fresno, Calif.-based gym is using AI-powered workout equipment that automatically adapts to the exercise routines of different users by sensing their maximum resistance to different weights  and ranges of motion, making instant adjustments. (The Business Journal) 

Facebook Inc. reached a deal with the Australian government to restore news pages to the social media company’s platform, following a five-day suspension because of a disagreement over payment for content. (The Wall Street Journal)

Dominion Voting Systems sued Mike Lindell, chief executive of MyPillow Inc. and a prominent supporter of former President Donald Trump, alleging that the businessman had defamed the company with false accusations that it had rigged the 2020 election for President Biden. (The Wall Street Journal)

A new Treasury Department watchdog report warns that law-enforcement agencies may not be on firm legal footing when they use cellphone GPS data drawn from mobile apps without obtaining a warrant first. (The Wall Street Journal)

A group of congressional lawmakers is renewing its call for a cyber ambassador in the State Department, reviving a bill that created friction between Congress and the Trump administration. (The Wall Street Journal) 

Write to Enterprise Technology Editor Steven Rosenbush and Pro AI Deputy Editor Angus Loten.

Follow us on Twitter: @wsjproai, @steve_rosenbush, @J_J_McCormick, @tomloftus, @SCastellWSJ, @JaredCouncil  and @angusloten.