Radical Reads

Trust and AI in Healthcare

By Sanjana Basu, Investor

In the midst of COVID-19, healthcare AI funding reached a record high of ~$6.6B in 2020. Investments fell into two primary categories: AI solutions that treat and diagnose diseases, and those that drive efficiencies in healthcare organizations. While we recognize the potential AI has to transform healthcare, patient trust remains a critical factor in the adoption of AI solutions by healthcare stakeholders.

Last week, the Consumer Technology Association (CTA) unveiled a new standard identifying core requirements for AI solutions to be deemed trustworthy. The standard identifies three major expressions of trust:

  1. Human trust – how humans perceive and interact with AI solutions including patient experience and explainability. Developers of AI solutions need to make clear what the solution can and cannot do, how the solution performs and put its abilities into context.
  2. Technical trust – promotes responsible use of data, focusing on data quality and requirements to eliminate bias. This section also addresses data access, privacy and security, all of which are essential for effective technical execution.
  3. Regulatory trust – which is maintained by complying with a regulatory system that ensures safety, effectiveness and reliability of healthcare AI products.

As of this week, 64 healthcare organizations have endorsed these standards. Earlier this year we wrote about FDA’s Action Plan to regulate Artificial Intelligence and Machine Learning in Software as Medical Devices. We are encouraged by additional programs aimed at increasing trust in AI solutions. Innovations in this technology remain essential to improving patient outcomes and reducing healthcare costs worldwide.

AI News This Week

  • AI Here, There, Everywhere  (New York Times)

    Daily interactions with artificial intelligence systems are expected to become increasingly personalized according to researchers. From kitchen appliances to garbage pickup, AI will soon intersect with increasing frequency with our everyday lives. As an example, computer vision pioneer Fei-Fei Li – who is also the co-founder of DawnLight, a Radical portfolio company – describes how AI systems will soon be able to support seniors living alone and catch early patterns of dementia, sleep disorders, social isolation, falls, and poor nutrition.

  • US Holds Slim Edge over China in Artificial Intelligence, Former Google Chairman Says  (USNI News)

    Former Google Chairman and CEO, Eric Schmidt, warned Congress that the United States is only two years ahead of China in developing AI. In his role as chairman of a special commission on AI, Schmidt says the United States needs to maintain a five to 10-year advantage over its competitor in AI and other high technology fields like quantum computing. The commission’s report is scheduled to be released this week. In an episode of Radical Talks last spring, Eric Schmidt sat down with our  Managing Partner, Jordan Jacobs, to discuss US-China relations, AI and healthcare, and more. Eric Schmidt is an investor in Radical Ventures.

  • Tech Firms Train Voice Assistants to Understand Atypical Speech  (Wall Street Journal)

    Voice assistants like Alexa and Siri cannot understand people with speech disorders, but their creators say they are looking to change that. Approximately 7.5 million people in the US have trouble using their voice and that group is at risk of being left behind by voice-recognition technology. Training voice assistants to respond to people with speech disabilities may also improve the experience of voice-recognition tools for seniors, a growing user base of voice technology.

  • Canadian AI Startup Presents PetaOPS Card  (EE Times)

    The inference chip is the engine of AI computation. The next-generation of AI breakthroughs will depend on hardware that can support AI systems at the edge, powering self-driving cars, smart cities, and vision guided robotics. A new chip architecture for neural net inference developed by Untether AI, addresses many of the biggest challenges for inference chips, dramatically increasing performance while reducing power consumption. Radical Ventures is an investor in Untether AI.

  • AI conquers challenge of 1980s platform games  (BBC News)

    Scientists have come up with a computer program that can master a variety of 1980s exploration games, paving the way for more self-sufficient robots. Previously, these scrolling platform games have been challenging to solve using artificial intelligence. The algorithms could help robots better navigate real-world environments.

Radical Reads is edited by Leah Morris (Senior Director, Velocity Program, Radical Ventures).