When it comes to embracing generative AI, healthcare is something of a straggler. Some 89% of organizations in pharma/healthcare have neither deployed nor experimented with the technology.

So reports a team of researchers at MIT Technology Review Insights. The authors came by the finding after conducting a global poll of 1,000 executives in multiple industries. They also interviewed experts in academics, data science and AI.

But generative AI participation is just an aside to the investigators’ primary interest. What they’re really after is what it takes to scale the technology from a toe in the water by one or two brave departments to a full-on swim by a broad swath of the enterprise.  

Sooner or later, healthcare and fellow slowpokes—most notably government, transport/logistics, energy/utilities and travel/hospitality—will take that plunge. In the meantime, here’s what they can do now to ensure, in the MIT authors’ words, “smooth scaling.”

1. Know that partnerships with both AI startups and Big Tech will be critical to smooth scaling.

“Most executives (75%) plan to work with partners to bring generative AI to their organization at scale. While Big Tech has an ecosystem advantage, startups enjoy advantages in several specialized niches. Executives are somewhat more likely to plan to team up with small AI-focused companies (43%) than large tech firms (32%).”

2. Pace for success.

“Firms should move steadily and cautiously to enterprise-wide AI deployment, especially in light of evolving regulatory and legal regimes. History shows that new technology adoption is often slowest at the start but can accelerate exponentially thereafter.”

3. Avoid overpredicting workforce reductions.

“Today’s decisions could spread worker productivity and value. And dystopian unemployment forecasts may overestimate the power of automation and undervalue the know-how of employees. Executives should look to the potential of technology to enhance customer and employee experience, not merely to cut costs.”

4. Allow time for new laws and oversight bodies to take shape.

“New legislation and institutions are warranted to manage AI risk, but these will take time to establish. In what may be AI’s ‘Napster moment,’ the current period could give rise to new business models and consumer norms.”

MIT produced the report with backing from several tech companies. Read the whole thing for free here.

Buzzworthy developments of the past few days.

• Generative AI is helping birth new unicorns at a brisk pace. Unicorns are private startups valued at $1 billion or more. Over the past 12 months, more than 60% of startups breaking through $1B barrier in the cloud and SaaS space have been GenAI-native. That’s according to the venture firm Accel, which released a report documenting the market movement Oct. 17. The data also show that valuations, investment levels and unicorn creation have returned to pre-COVID levels. However, the business environment has changed a lot since 2019, Accel Euroscape partner Philippe Botteri remarks. “Companies have moved away from growth at all costs to focus on profitability,” he says. And GenAI is “redefining the potential of software, bringing the opportunities for enterprise automation and productivity improvement to a new level.” Access the full report here.
   
• FDA is forming a new committee to guide its views on digital health technologies. In its sights will be not just AI but also augmented reality, virtual reality, wearables, remote patient monitoring—all that sort of stuff. The Digital Health Advisory Committee’s marching orders will be to help “improve the agency’s understanding of the benefits, risks and clinical outcomes” associated with digital health technologies. The new arm is slated to flex its budding bicep sometime next year. Details here.
   
• AI has achieved 100% accuracy at identifying melanoma. The same software detected 99.5% of all skin cancers and 92.5% of precancerous lesions. This performance marked a significant improvement over the first time that particular AI model was tested, which was only two years ago. All that’s left for it now is to hit 100% accuracy across all three diagnoses. More on the latest research here.    
   
• Only 4% to 5% of enterprises presently employ tech experts dedicated to things like AI strategy and robot management. Those kinds of roles will have to multiply if organizations are to leverage emerging technologies for use cases that add real value. So suggests Gartner analyst Lily Mok in commentary published in Information Week. The piece is mainly aimed at CIOs but should apply to anyone interested in the latest thinking on “how to build a resilient IT workforce with AI and unconventional talent.”
   
• ‘AI could spur an economic boom. Humans are in the way.’ And there’s AIin.Healthcare’s favorite AI headline of the week. It’s from the Wall Street Journal. In the body, economics reporter Amara Omeokwe suggests patience will be needed by those anxious to wring hard ROI from AI. “[I]t could take time for AI’s effects to actually show up in productivity measures,” she writes. “[C]ompanies will need to hire or train specialized workers who understand both generative AI’s capabilities and enough about existing production processes so that the two can be linked.” Where have we heard something like that before? (See item immediately above.)
   
• Siemens Healthineers is teaming with an international health organization to bring tuberculosis to heel. The German healthcare giant and the Global Fund to Fight AIDS, Tuberculosis and Malaria will focus on using AI to help screen for and diagnose TB in regions short on human expertise. First up is Indonesia, where TB has been spreading fast and less than half of infected people have been receiving treatment. Joint announcement here.
   
• Epic has been quietly incorporating generative AI across the domain of healthcare technology that it knows best. That would be electronic health records. Catch up on how far the company has taken the tech in a 1-minute video posted here.
   
• Samsung subsidiary Harman has come out with a large language AI model just for healthcare. Calling the product HealthGPT, Harman says the product can help with clinical decision-making, medical dataset customization, cost control and drug discovery. Announcement.
   
• Learning & development professionals working in healthcare are looking to generative AI with great expectations. Some 88% believe the technology has at least moderate potential for training healthcare workers, while substantial subgroups think its potential is high or even “very high.” The findings are from a survey conducted by L&D technology company Virti. More results here.
   
• From AIin.Healthcare’s news partners:
   
  • Radiology Business: Number of FDA-cleared AI imaging tools expected to skyrocket by 2035, backed by $30B in funding
  • Cardiovascular Business: FDA clears GE Healthcare’s new image-guided system for heart patients of all sizes