An interview with J. Craig Wheeler, Professor of astronomy of the University of Texas at Austin about his book, The Path to Singularity.
Explore how synthetic systems merge with organic experience in The Artificiality. Follow the journey from information to consciousness, where AI evolves from passive tools to active participants, reshaping reality, redefining intelligence, and challenging the boundaries of human experience.
Congratulations to the authors of of the Artificiality Book Awards 2024!
The math of COVID-19 may mean some level of opt-in tracking is vital to stop repeated outbreaks
Preventing transmission is a top priority in the fight against SARS-CoV-2. Right now, there are two strategies for this: social distancing/widespread lock-downs and contact tracing/case isolation. While not a true tradeoff, there is an inverse relationship between having a granular understanding of, and response to, the spread and the need for blunt-force wide-spread quarantine measures.
There are many initiatives that provide movement data from publicly available information — the NYT, Unacast and Google to name a few. But these systems only show how much social distancing is slowing people down. They show nothing about how slowing people slows disease. This is where testing and contact tracing comes in.
Traditional contact tracing and case isolation is highly manual. The process of tracking down people who have been exposed to the virus takes three days per case, which is an impossible hurdle in the US given low numbers of public health workers and the high numbers of new cases per day, not to mention that contact tracing can be highly privacy invasive, especially when the government enforces quarantine.
Understandably, people are turning to digital solutions.
However, Americans are wary of giving up their data for this purpose. In a recent survey, only a third of people said they’d share publicly available travel information and even fewer were willing to have their location tracked via cellphone, either on an individual basis or in aggregate. People are also cautious about who they would share the data with. 55% said that they’d want a positive COVID-19 test shared with public health authorities such as WHO or CDC, compared with 35% and 27% for local and federal government. Trust in app developers, social media or start ups is very low:
Just 21% would want a positive result shared with an app that discloses where people are infected through anonymized or aggregated data, while 9% would share it with an app that discloses the names of people infected. - Axios
But we might need these apps. The math of this pandemic makes digital surveillance too valuable to ignore because the virus spreads too fast for traditional contact tracing to be effective. This is according to new research from the University of Oxford.
A key measure of transmission is R0 (pronounced R naught) - the typical number of infections caused by an individual when no one has immunity. Once R0 is less than 1, the epidemic fizzles out. One of the key problems with this novel virus is that it appears there is a significant level of transmission from pre-symptomatic people — the researchers suggest between a third and a half of transmissions. Pre-symptomatic cases have an estimated R0 of 0.9, which the researchers say is almost enough to sustain the epidemic on its own. (Estimates for overall R0 are somewhere between 1.5 and 3.5.)
So even if we could effectively manually contact people in a privacy-protective way, it wouldn’t be enough to reduce the R0 to less than 1, so it wouldn’t be enough to stop the spread.
But speed can substitute for effectiveness, which is where digital tracking comes in. Instead of 80% of cases needing to be found and isolated, and 80% of their contacts of their contacts quarantined, instantaneous digital tracing could reduce this to only 60% of cases needing to be isolated and 60% of contacts traced — and possibly as few as 50% and 40%, respectively.
A Seattle infectious disease modeling expert, Trevor Bedford, is developing an app called NextTrace that aims to do this and hopes to get enough opt-in users that it can be effective in stalling the spread.
NextTrace therefore plans to build a decentralized reporting system in which anyone with confirmed Covid-19 can choose to register, anonymously, on an online platform. The platform will use cellphone location and proximity data from cellphones, for people who have opted in, to find individuals who might have been exposed to this case and advise them to be tested. The system would build a contact history for each case. - StatNews
This makes all sorts of sense. Bedford probably would count as a “public health agency” in the eyes of the public and if the system can truly work anonymously then enough people may sign up that tracing/self-isolation can reach the threshold required. The rub will be whether people who are flagged as COVID risks then do the responsible thing and self-isolate. This next step is obviously vital and without that assurance we are back to asking “what role does the government have in enforcing quarantine?”
AI ethics experts are right to point out the privacy perils which arise in a pandemic. Renee Cummings, Founder & CEO Urban AI, in an online conversation this week said that “AI can make a difference in real-time” but many countries are taking draconian approaches to contact tracing and monitoring. “While some of these might be very effective, what we are seeing is systems integration between public health and national security. Many of the strategies that are being used are counter-terrorism, law enforcement and national security strategies.” When we are through the other side, we may realize that many of our civil liberties have gone in the effort to flatten the curve.
What can we do about this in the middle of the crisis? As individuals, not much, said Reid Blackman, CEO at Virtue, in the same conversation. But as a collective we can demand that companies are transparent and commit to measures such as not sharing or selling this data, only using the data for the purposes of fighting the pandemic, not sharing data with government agencies unrelated to public health, returning or deleting the data when this is done and not using data for commercial gain.
Perhaps the real reason that some form of digital surveillance is inevitable is that this epidemic could come and go in waves for quite some time, which means granular and targeted ways to deal with outbreaks are too valuable to ignore, especially as the US seems unable to coordinate a lock-down response.
It’s not too late to plan for outbreaks in the fall or winter, to get ahead of these basic privacy questions and secure or clarify individual rights (say, a compulsory quarantine order versus trust).
It’s conceivable that the entire country will have to go through this crisis multiple times before a vaccine or effective treatment arrives so we should be doing more to prepare privacy protections in advance. In this way, enough people might opt for digital surveillance and alerts which could minimize future broad-swath shelter-in-place directives.
Also this week:
Writing and Conversations About AI (Not Written by AI)
