[This is a linkpost to Quantum Computing : A preliminary research analysis report]
Quite recently, Google released a paper claiming to have build a programmable quantum computer capable of solving in 20 milliseconds a sampling problem that would take 2.5 days to solve in the fastest 2019 supercomputer.
Among its potential uses, Quantum Computing (QC) will allow breaking classical cryptographic codes, simulate large quantum systems and faster search and optimization.
This could have implications on some of the areas of interest to a long termist, including in Artificial Intelligence, Biosecurity and Atomically Precise Manufactoring - as well as presenting new technological risks (such as undermining the current infrastructure for online transactions).
In response to the recent developments in the field and the above considerations, I resolved to discuss the relevance of Quantum Computing (QC) from the point of view of a philanthropist, as to better understand the risks and benefits posed by this technology.
A summary of the main conclusions I reached is reproduced below. You can see the full report here. Suggestions and comments are welcome!
- QC could unlock a significant amount of computing power for quantizable algorithms, and this could have implications on AI timelines and the kind of AI designs we will first see. This area of research seems well funded by big companies (Google, IBM, Microsoft). More
- Due to its potential applications to advance transformative AI, I would be medium excited to see further research on the effects of QC compute overhang on AI timelines and possible policy implications for regulation of AI development. This line of research seems however bottlenecked by our understanding of how extra compute in general affects AI timelines, research on QC timelines and research in QC policy. More
- I expect QC developments neither hinder nor help with bottlenecks in current research agendas in AI alignment, and I see no need to currently invest in expertise in this intersection. More
- QC compromises secure communication and online transactions, which could have profound economic effects. The scale of this problem is unclear. This issue seems to be receiving a commensurate amount of attention from government orgs (NIST) and industry (ETSI), and overall tractable. More
- QC could have an effect on computer hardware design. I am unsure of how promising this research line is compared to other research lines in hardware design. I’d be keen on seeing a 10 hour research project reflecting on this. More
- Very tentatively, I’d guess that the development of QC would not significantly increase bio risk nor it is a specially promising tool for mitigation. More
- QC would likely not have a major effect on Atomically Precise Manufacturing. Current research in APM is able to abstract away from the quantum mechanics of precise protein interactions and thus able to use efficient classical simulations for design. More
- QC has some promising applications with little apparent downside in medicine, agriculture (fertilizer design) and operations research. Big companies like IBM, Google and Microsoft are aware of these applications, so I would expect little room-for-funding. It is also unclear whether there is lower hanging fruit for these applications via other avenues. More
- Some researchers are preemptively exploring QC simulations as a basic research tool in physics and chemistry, and research of QC limits as physical limits. However it is not clear to me that basic research on physics is urgent, and I am also uncertain about how useful QC tools would be in chemistry research. An interview with an expert in quantum chemistry may clarify the issue. More
- Some strategic and tractable research on general Quantum Computing that would be helpful to better understand its relevance includes 1) forecasting QC timelines and 2) researching whether and how QC can be regulated. I think that a 100h research project on those questions is worth funding insofar it will allow us to better understand the VOI of strategic research on concrete applications and regulation of QC. More
- I would recommend to fund some more exploratory research in QC grantmaking as a high-risk high-reward research project. I would however recommend against funding object level research or pursuing careers in QC for philanthropic reasons. More
In this summary I have given guesses on what is worth funding. Researchers may want to look into the sections they are considering investigating themselves for open questions.
This article was written by Jaime Sevilla. This work was partially supported by the Future of Humanity Institute summer fellowship program and partly by a grant made by the Effective Altruism Foundation.
I want to thank Pablo Moreno for discussion on Quantum Computing, Jassi Pannu and Gregory Lewis for discussion on biological risks, Eric Drexler for discussion on Atomically Precise Manufacturing, Max Daniel for mentorship and Luisa Rodriguez for general feedback.