|
|
|
Image by Christian Ohde/IMAGO via Reuters
|
|
AI is already being used to optimize supply chains, advance the development of pharmaceutical drugs, and pilot fighter jets. But despite the promise and the hype, many organizations are struggling to deliver working AI applications. In fact, by some estimates, more than 80 percent of AI projects fail. That’s twice the rate of failure for IT projects that do not involve AI.
To understand why so many AI projects fall short, RAND researchers interviewed 65 data scientists and engineers with at least five years of experience in building AI or machine learning models. This revealed five root causes of AI failures:
- Industry stakeholders often misunderstand—or miscommunicate—what problem needs to be solved using AI.
- Many organizations lack the data to adequately train an effective AI model.
- There’s too much focus on using new technology—and too little focus on using AI to solve real problems.
- Some organizations might lack adequate infrastructure to manage their data and deploy AI models.
- AI technology is sometimes applied to problems that are simply too difficult for it to solve.
To minimize these issues, organizations can ensure that technical staff understand the purpose of AI projects; invest in robust infrastructure; and work to understand AI's limitations. It's also essential to focus on the problem—not the technology. Chasing the latest and greatest AI models for their own sake, RAND researchers write, is one of the most frequent pathways to failure.
|
|
|
|
|
|
|
|
|
The Wireless Emergency Alerts system sends public safety alerts, AMBER alerts, and imminent threat alerts to Americans' phones. RAND researchers recently surveyed U.S. adults immediately after a nationwide test of this system. They found that most people with working cell phones received the test alert. However, receipt of the alert varied based on certain factors, such as phone carrier, age, and community demographics. Further, one in six respondents said they opted out of at least one type of alert. These findings point to a few key focus areas that could help improve the reach of important public safety alerts. Read more »
|
|
|
|
Flooding has become the most expensive and frequent disaster in the United States. Flooding events occur almost daily, causing not just physical damages, but also degrading housing conditions, limiting economic opportunities, and harming ecosystems—all of which affect human health and well-being. According to RAND experts, building flood resilience requires transforming systems that are fundamental to how and where people live, work, and build. This includes taking innovative approaches to data collection, finding new ways to design community infrastructure, and focusing on collaboration among existing governing structures and financing mechanisms. Read more »
|
|
|
|
|
|
RAND Recommends
|
|
|
- RAND's Mark Schwartz spoke to CBS News about the threat of an Iranian strike on Israel and the potential for a political solution to end the Israel-Hamas war.
- RAND researchers Elina Treyger and Douglas Ligor spoke to NPR about last year's surge in Chinese migrants at the U.S. southern border.
- “There almost is no relationship between the U.S. and Russia,” RAND's Samuel Charap told The World. “The climate is one of deep antagonism, so the arms control architecture has been in a period of decline and almost collapsed over the last 20-some years.”
|
|
|
|
|
|
Events
|
|
|
|
|
Friday, August 16, 2024 – Online (More dates available at the link above.)
|
|
|
Thursday, August 29, 2024 – Online
|
|
|
|
|
|
