Home Money National security expert crunches numbers on dangers of AI

National security expert crunches numbers on dangers of AI

0 comment
National security expert crunches numbers on dangers of AI

One type of risk that has been of great interest to him for a long time is “biohazard.” which is the worst What could happen? Take us through that.

I started in public health before working in national security, working on infectious disease control: malaria and tuberculosis. In 2002, the first virus was synthesized from scratch on a Darpa project, and it was kind of an “oh shit” moment for the biosciences and public health community, realizing that biology is going to become an engineering discipline that could potentially be misused. He was working with veterans of the smallpox eradication campaign and they thought, “Holy shit, we just spent decades eradicating a disease that could now be synthesized from scratch.”

I then went on to work in biosecurity, trying to figure out how we could increase security around biolabs so they would be less likely to be used. How can we detect biological weapons programs? Which, unfortunately, still exist in significant quantities in some places around the world. Also, how can we build more safety into society so that we are more resilient when it comes to not only an engineered pandemic but also natural pandemics?

There is a lot of vulnerability that remains in society. Covid was a demonstration of this. This was a relatively mild virus by historical standards (it had an infection fatality rate of less than 1 percent), while there are some naturally occurring viruses that have fatality rates well above 50 percent. There are synthetic viruses that have a lethality close to 100 percent and, at the same time, are as transmissible as SARS-CoV-2. Although we know how to design vaccines and manufacture them very quickly, getting them approved takes about as long today as it did about 20 years ago. So the amount of time it would take to vaccinate a population is about the same today as it was for our parents and even our grandparents.

When I became interested in biosecurity in 2002, building a poliovirus, a very, very small virus, cost many millions of dollars. It would have cost about a billion dollars to synthesize a smallpox virus, a very large virus. Today, the cost is less than $100,000, so that’s a 10,000-fold decrease over that period. Meanwhile, the cost of vaccines has actually tripled over that period. The defense-offensive asymmetry is moving in the wrong direction.

And who do you think is our greatest adversary in terms of biological risks?

Nature comes first. The evolution of natural viruses continues. We are going to have future viral pandemics. Some of them will be worse than Covid, others will not be as bad as Covid, but we have to be resilient to both. Covid cost the US economy alone more than 10 billion dollarsAnd yet, what we invest in preventing the next pandemic is perhaps $2 billion to $3 billion in federal investment.

Another category is intentional biological attacks. Aum Shinrikyo was an apocalyptic sect in Japan that had a biological weapons program. They believed they would fulfill the prophecy by killing everyone on the planet. Fortunately, they were working with 1990s biology, which wasn’t as sophisticated. Unfortunately, they then turned to chemical weapons and launched the sarin gas attacks on Tokyo.

Today we have individuals and groups intent on mass casualties expressing increasing interest in biology as a weapon. What prevents them from using biology effectively is not controls over tools or raw materials, because all of that is now available in many laboratories and on eBay; You can now buy a DNA synthesizer for much less than $100,000. You can get all the materials and consumables you need at most scientific supply stores.

What an apocalyptic group would lack is the knowledge to turn those tools into a biological weapon. There is concern that AI will make knowledge more available. Some of the investigation made by [AI safety and research company] Anthropic has analyzed risk assessments to see if these tools could be misused by someone without a strong biological background. Could they basically get postgraduate training from a digital tutor in the form of a large language model? Right now probably not. But if you look at the progress of recent years, the barrier to entry for someone who wants to carry out a biological attack is eroding.

You may also like