What role does AI play in responding to a pandemic? The National Security Commission for AI provides a framework


(via Fotolia.com)

Earlier this year, I attended the third annual AI Summit at the Potomac Officers Club, a conference on AI for defense and intelligence. At the event, Executive Director Yii Bajaktari of the National Security Commission on Artificial Intelligence (NSCAI) shared the highlights of his Magnum Opus Final Report National Security Commission on Artificial Intelligence.

A huge 750-page piece carried an unpleasant message: America is not ready to defend or compete in the age of AI. I have checked the content of the report here.

NSCAI recently published several white papers, one of which was also written by Dr. Bajaktari was written. It’s a far more pleasant report, titled “The Role of AI Technology in Pandemic Response and Preparedness: Recommended Investments and Initiatives,” at just under thirty pages. Since this whitepaper contains some interesting and potentially new ideas, I’ll try to highlight and summarize some of them.

The introduction was pretty lukewarm and didn’t introduce any new concepts:

While it is unclear whether AI will significantly change the course of COVID-19, this initial research shows how AI could be used to improve pandemic preparedness and response in the future. With proper investment over the next decade, AI could revolutionize how scientists understand data, conduct research and develop new drugs, how doctors diagnose certain diseases and interact with their patients, and how public health officials manage information and make decisions.

We know that AI is not a panacea and that some technical promises are more theoretical than practical today. Algorithms can work differently in the field than in the laboratory. Bias can creep into the underlying data and model design. Many applications remain brittle with little portability or effectiveness outside of their narrow and discrete use cases. In order to teach doctors and scientists to use new tools and to integrate them into their work processes, education and training are required. Care must be taken to ensure that the systems and tools used are secure, trustworthy, impartial and understood by the users.

The topics of the first part of the report are largely ambitious or unquantified initiatives in scope, but part two focuses on funding initiatives:

Analysis of part one

In Part One, “Promising Uses of AI in Pandemic Response – Now and in the Future,” the report says:

Situation awareness and disease monitoring: AI is used to better understand the spread of disease. For example, Boston-based AI company DataRobot has developed an AI model to predict the spread of COVID-19 down to the county level. Canadian company BlueDot recognized the spread in Wuhan long before it was taken seriously.

Today’s AI capabilities can already be used to sift through data to identify vulnerabilities and zoonotic spillovers (the transmission of pathogens from a vertebrate to a human). Though preliminary, recent data suggests that bats are the most likely source of origin for the current 2019-novel CoV (2019nCoV.) Outbreak, which began in Wuhan, China in December 2019). AI will advance bioforensics by enabling inference and predictive models. And likely to play an increasing role in the identification of new sensor materials as well.

I see this mainly as something worth striving for. Being able to diagnose a problem seems like a good step, but there are already many failures out there. The health system is profit and performance-driven. I doubt this will have any impact.

My Opinion on Vaccines, Therapeutics and Medical Devices: It has already been shown that AI was instrumental in the rapid development of vaccines against COVID-19. However, many of these had been in development for years. Pharmaceutical companies have redirected their research to this one topic, so the “miracle” of COVID-19 vaccines wasn’t quite that.

However, the claim in this paper that AI will facilitate the reuse of existing therapies for novel applications through simulation is a promising idea and already in practice. AI to drive supply chains and advanced manufacturing requires far more inventions and resources. While AI can add some capabilities to complex supply chains, the discipline is much more complicated. I wrote about that here.

With regard to Ongoing Health Care – Protecting the Healthy and Treating the Sick, I reviewed an AI model that is widely used in hospitals and has serious ethical and functional problems. Still, healthcare is ripe for a generation leap with AI, but problems are waiting for it.

Analysis of part two

The report’s “Recommended AI-Focused Investments and Initiatives” provide context for the first part. In particular, the need for government funding. In “Using AI to Advance Science”:

A. Doubling federal funding for non-defense AI research and development to approximately $ 2 billion to use AI for future pandemic response. Novel machine learning instructions and test evaluation verification and validation (TEVV). NSF $ 400 million, DOE $ 300, NIH $ 150, and NIST $ 50 million.

B. Create an NSF-run AI health and biomedicine institute. Six $ 4 million educational grants to advance basic research.

C. Launch a price challenge managed by NSF to drive next generation data integration and modeling. Advances in AI reasoning could help expand SIR modeling (A SIR model is an epidemiological model that calculates the theoretical number of people who have become infected with a contagious disease over time in a closed population ).

This seems a little strange given that Congress has already passed the United States Innovation and Competition Act of 2021, formerly known as the Endless Frontier Act, which allocates $ 110 billion to basic and cutting-edge technology research over five years and a complete reorganization approved the NSF. Politics got in the way, however, and although passed down bipartisan, it diluted funding by about half. The bill was supposed to demonstrate US commitment to compete with China in AI, but the message is now mixed.

The report continues:

D. Continuation of DARPA’s work to further develop the architecture for data exchange and collaboration in drug discovery and mirroring efforts at the NIH.

This is a nice idea, but the top performing supercomputers in the National Labs have air gaps, which means they are not connected to anything, so real-time data exchange is not likely.

e. Make the COVID-19 high performance computer consortium permanent.

As mentioned earlier, there are national security problems. Many of the largest computers are involved in nuclear weapons research.

From the section “Building an AI-enabled basis for smart response”:

A. Create a state pandemic response dataset.

This seems like a galactic effort unless it is tightly focused. On the commercial side, C3.ai built a COVID-19 data lake that is open to the research community.

B. Invest in digitally upgrading state and local health infrastructure for effective disease surveillance.

Interesting idea, but states disagree. Would states that threaten to arrest masked teachers cooperate?

C. Enhance global cooperation in intelligent disease surveillance and international health norms and standards.

I don’t see this being primarily an AI driven policy.

D. Invest in AI-powered skills to maintain military readiness.

There are many details in the report surrounding the simple premise of using the methods described above to keep the military safe from disease.

My recording

I find all of these really admirable ideas. But it seems that any move by Congress, just like the US Innovation and Competition Act of 2021, is doomed to partisan bickering.


Comments are closed.