2019-20 New Year review

This is an annual post reviewing the last year and making resolutions and predictions for next year. This year’s edition features sleep tracking, intermittent fasting, overcommitment busting, and evaluating calibration for all annual predictions since 2014.

2019 review

AI safety research:

• Wrote an updated version of the relative reachability paper, including an ablation study on design choices.
• Coauthored a paper on modeling AGI Safety frameworks with causal influence diagrams, accepted to IJCAI workshop.
• Wrote a paper on avoiding side effects by considering future tasks, accepted to NeurIPS workshop.
• Co-ran a subteam of the safety team focusing on agent incentive design (ongoing).

AI safety outreach:

• Co-organized FLI’s Beneficial AGI conference in Puerto Rico, a more long-term focused sequel to the original Puerto Rico conference and the Asilomar conference. This year I was the program chair for the technical safety track of the conference.
• Co-organized the ICLR AI safety workshop, Safe Machine Learning: Specification, Robustness and Assurance. This was my first time running a paper reviewing process.
• Gave a talk at the IJCAI AI safety workshop on specification, robustness an assurance problems.
• Took part in the DeepMind podcast episode on AI safety (“I, robot”).

Work effectiveness:

• At the beginning of the year, I found myself overcommitted and kind of burned out. My previous efforts to reduce overcommitment had proved insufficient to not feel stressed and overwhelmed most of the time.
• In February, I made a rule for myself to decline all non-research commitments that don’t seem like exceptional opportunities. The form that I made last year for evaluating commitments (which I have to fill out before accepting anything) has been helpful for enforcing this rule and avoiding impulsive decisions. The number of commitments went down from 24 in 2019 to 10 in 2019. This has been working well in terms of having more time for research and feeling better about life.
• Organizing a conference and a workshop back to back was a bit much, and I feel done with organizing large events for a while.

• Stopped using work cycles and pomodoros since I’ve been finding the structure a bit too restrictive. Might resume at some point.
• Hours of “deep work” per month, as defined in the Deep Work book. This includes things like thinking about research problem, coding, reading and writing papers. It does not include email, organizing, most meetings, coding logistics (e.g. setup or running experiments), etc.

• For comparison, deep work hours from 2018. My definition of deep work has somewhat broadened over time, but not enough to account for this difference.

Health / self-care:

• I had 7 colds this past year, which is a lot more than my usual rate of 1-2 per year. The first three were in Jan-Feb, which seemed related to the overcommitment burnout. Hopefully supplementing zinc will help.
• Averaged 7.2 hours of sleep per night, excluding jetlag (compared to 6.9 hours in 2018).
• About a 10% rate of insomnia (excluding jetlag), similar to the end of last year.
• Tried the Oura ring and the Dreem headband for measuring sleep quality. The Oura ring consistently thinks I wake up many times per night (probably because I move around a lot) and predicts less than half an hour each of deep and REM sleep. The Dreem, which actually measures EEG signals, estimates that I get an average of 1.3 hours of deep sleep and 1.8 hours of REM sleep per night, which is more than I expected.
• Started a relaxed form of intermittent fasting in March (aiming for a 10 hour eating window), mostly for longevity and to improve my circadian rhythm. My average eating window length over the year was 10.5 hours, so I wasn’t very strict about it (mostly just avoiding snacks after dinner). One surprising thing I learned was that I can fall asleep just fine while hungry, and am often less hungry when I wake up. My average hours of sleep went up from 6.96 in the 6 months before starting intermittent fasting to 7.32 in the 6 months after. I went to sleep 44 minutes earlier and waking up 20 minutes earlier on average, though the variance of my bedtime actually went up a bit. Overall it seems plausibly useful easy enough to continue next year.

Fun stuff:

• Did a Caucasus hiking trek in Georgia with family, and consumed a lot of wild berries and hazelnuts along the way.

• Did a road trip in southern Iceland (also with family), saw a ridiculous number of pretty waterfalls, and was in the same room with (artificial) lava.

• Took an advanced class in aerial silks for the first time (I felt a bit underqualified, but learned a lot of fun moves).
• Ran a half-marathon along the coast in Devon on hilly terrain in 3 hours and 23 minutes.
• Made some progress on handstands in yoga class (can hold it away from the wall for a few seconds)
• Did two circling retreats (relational meditation)
• Read books: The Divide, 21 Lessons for the 21st Century, The Circadian Code, So Good They Can’t Ignore You, Ending Aging (skimmed).
• Got into Duolingo (brushed up my Spanish and learned a bit of Mandarin). Currently in a quasi-competition with Janos for studying each other’s languages.

2019 prediction outcomes

Resolutions:

1. Author or coauthor two or more academic papers (50%) – yes (3 papers)
2. Accept at most 17 non-research commitments (24 last year) (60%) – yes (10 commitments)
3. Meditate on at least 250 days (60%) – yes (290 days)

Predictions:

• Relative reachability paper accepted at a major conference, not counting workshops (60%) – no
• Continue avoiding processed sugar for the next year (85%) – no (still have the intention and mostly follow it, but less strictly / consistently)
• 1-2 housemate turnover at Deep End (2 last year) (80%) – yes (1 housemate moved in)
• At least 5 rationality sessions will be hosted at Deep End (80%) – no

Calibration over all annual reviews:

• 50-70% well-calibrated, 80-90% overconfident (66 predictions total)
• Calibration is generally better in 2017-19 (23 predictions) than in 2014-16 (43 predictions). There were only 3 70% predictions in 2017-19, so the 100% accuracy is noisy.
• Unsurprisingly, resolutions are more often correct than other predictions (72% vs 56% correct)

2020 resolutions and predictions

Resolutions

• Author or coauthor three or more academic papers (3 last year) (70%)
• At most 12 non-research commitments (10 last year) (80%)
• Meditate on at least 270 days (290 last year) (80%)
• Read at least 7 books (5 last year) (70%)
• At least 700 deep work hours (551 last year) (70%)

Predictions

• I will write at least 5 blog posts (60%)
• Eating window at most 11 hours on at least 240 days (228 last year) (70%)
• I will visit at least 4 new cities with population over 100,000 (11 last year) (70%)
• At most 1 housemate turnover at Deep End (70%)
• I finish a language in Duolingo (60%)

Past new year reviews: 2018-19, 2017-18, 2016-17, 2015-16, 2014-15.

 

2018-19 New Year review

2018 progress

Research / AI safety:

• Wrote a paper on measuring side effects using relative reachability in May, and presented the results at the ICML GoalsRL workshop and the AI safety summer school. Since then, some new approaches have come out using my method as a baseline :).
• Made a list of 30 specification gaming examples in AI (assembled from several existing lists). Since the list was posted in April, 16 new examples have been contributed through the form (thanks everyone!). The list received some attention on Twitter, and I was interviewed about it by Wired and the Times.
• Was in the top 30% of NeurIPS reviewers.
• Gave talks at the Oxford AI Society, EA Global London, etc.
• Got involved in organizing the upcoming ICLR AI safety workshop, Safe Machine Learning: Specification, Robustness and Assurance.

Rationality / effectiveness:

• Attended the CFAR mentoring workshop in Prague, and started running rationality training sessions with Janos at our group house.
• Started using work cycles – focused work blocks (e.g. pomodoros) with built-in reflection prompts. I think this has increased my productivity and focus to some degree. The prompt “how will I get started?” has been surprisingly helpful given its simplicity.
• Stopped eating processed sugar for health reasons at the end of 2017 and have been avoiding it ever since.
  • This has been surprisingly easy, especially compared to my earlier attempts to eat less sugar. I think there are two factors behind this: avoiding sugar made everything taste sweeter (so many things that used to taste good now seem inedibly sweet), and the mindset shift from “this is a luxury that I shouldn’t indulge in” to “this is not food”.
  • Unfortunately, I can’t make any conclusions about the effects on my mood variables because of some issues with my data recording process :(.
• Declining levels of insomnia (excluding jetlag):
  • 22% of nights in the first half of 2017, 16% in the second half of 2017, 16% in the first half of 2018, 10% in the second half of 2018.
  • This is probably an effect of the sleep CBT program I did in 2017, though avoiding sugar might be a factor as well.
• Made some progress on reducing non-research commitments (talks, reviewing, organizing, etc).
  • Set up some systems for this: a spreadsheet to keep track of requests to do things (with 0-3 ratings for workload and 0-2 ratings for regret) and a form to fill out whenever I’m thinking of accepting a commitment.
  • My overall acceptance rate for commitments has gone down a bit from 29% in 2017 to 24% in 2018. The average regret per commitment went down from 0.66 in 2017 to 0.53 in 2018.
  • However, since the number of requests has gone up, I ended up with more things to do overall: 12 commitments with a total of 23 units of workload in 2017 vs 19 commitments with a total of 33 units of workload in 2018. (1 unit of workload ~ 5 hours)

2017-18 New Year review

2017 progress

Research/career:

• Coauthored RL with reward corruption paper and presented the results at the U Toronto CS department, Workshop on Reliable AI, and Women in ML workshop.
• Coauthored AI Safety Gridworlds paper.
• Gave a talk on Interpretability for AI safety at the NIPS Interpretable ML Symposium.
• Gave a lot of people career advice on getting into AI safety research.

FLI / other AI safety:

• Coorganized the Beneficial AI conference in Asilomar and gave a talk summarizing the work of FLI grantees presented at the Asilomar workshop.
• Cowrote the project examples document for the new grants program.
• Spoke at the Tokyo AI & Society symposium (my first conference in Asia).
• Started a new public Facebook group AI Safety Open Discussion.

Takeaways from self-tracking data

I’ve been collecting data about myself on a daily basis for the past 3 years. Half a year ago, I switched from using 42goals (which I only remembered to fill out once every few days) to a Google form emailed to me daily (which I fill out consistently because I check email often). Now for the moment of truth – a correlation matrix!

The data consists of “mood variables” (anxiety, tiredness, and “zoneout” – how distracted / spacey I’m feeling), “action variables” (exercise and meditation) and sleep variables (hours of sleep, sleep start/end time, insomnia). There are 5 binary variables (meditation, exercise, evening/morning insomnia, headache) and the rest are ordinal or continuous. Almost all the variables have 6 months of data, except that I started tracking anxiety 5 months ago and zoneout 2 months ago.

2016-17 New Year review

2016 progress

Research / career:

• Got a job at DeepMind as a research scientist in AI safety.
• Presented MiniSPN paper at ICLR workshop.
• Finished RNN interpretability paper and presented at ICML and NIPS workshops.
• Attended the Deep Learning Summer School.
• Finished and defended PhD thesis.
• Moved to London and started working at DeepMind.

FLI:

• Talk and panel (moderator) at Effective Altruism Global X Boston
• Talk and panel at the Governance of Emerging Technologies conference at ASU
• Talk and panel at Brain Bar Budapest
• AI safety session at OpenAI unconference
• Talk and panel at Effective Altruism Global X Oxford
• Talk and panel at Cambridge Catastrophic Risk Conference run by CSER

Using humility to counteract shame

“Pride is not the opposite of shame, but its source. True humility is the only antidote to shame.”

Uncle Iroh, “Avatar: The Last Airbender”

 

Shame is one of the trickiest emotions to deal with. It is difficult to think about, not to mention discuss with others, and gives rise to insidious ugh fields and negative spirals. Shame often underlies other negative emotions without making itself apparent – anxiety or anger at yourself can be caused by unacknowledged shame about the possibility of failure. It can stack on top of other emotions – e.g. you start out feeling upset with someone, and end up being ashamed of yourself for feeling upset, and maybe even ashamed of feeling ashamed if meta-shame is your cup of tea. The most useful approach I have found against shame is invoking humility.

2015-16 New Year review

2015 progress

Research:

• Finished paper on the Selective Bayesian Forest Classifier algorithm
• Made an R package for SBFC (beta)
• Worked at Google on unsupervised learning for the Knowledge Graph with Moshe Looks during the summer (paper)
• Joined the HIPS research group at Harvard CS and started working with the awesome Finale Doshi-Velez
• Ratio of coding time to writing time was too high overall

FLI:

• Co-organized two meetings to brainstorm biotechnology risks
• Co-organized two Machine Learning Safety meetings
• Gave a talk at the Shaping Humanity’s Trajectory workshop at EA Global
• Helped organize NIPS symposium on societal impacts of AI

Rationality / effectiveness:

• Extensive use of FollowUpThen for sending reminders to future selves
• Mapped out my personal bottlenecks
• Sleep:
  • Tracked insomnia (26% of nights) and sleep time (average 1:30am, stayed up past 1am on 31% of nights)
  • Started working on sleep hygiene
  • Stopped using melatonin (found it ineffective)