A survey of hundreds of AI researchers by the AAAI finds that most don't think 'scaling up' nor neural nets alone will be sufficient to get us to human-level artificial intelligence.
The report, released at the annual meeting of the Association, calls for a broader focus of research on all kinds of AI, including 'symbolic AI', sometimes called 'good old fashioned AI'.
My story for Nature https://www.nature.com/articles/d41586-025-00649-4
We just had a 5.1 earthquake here in BC (fortunately it didn't do much damage; sadly I didn't get to feel it). This has sparked a lot of general questions on social media about quakes. As a science journalist who has written about earthquakes before, I thought I’d write up a quick FAQ here. This is just the result of my quick trawl through peer reviewed research and government websites; I'm not an expert, but I do a better job than AI, I like to think. :)
How many quakes do we normally get?
It is estimated that there are 500,000 detectable earthquakes in the world each year. 100,000 of those can be felt, and 100 of them cause damage. (USGS) In BC, there are several thousand quakes a year, of which about 50 are felt. A damage-causing quake happens on average once in BC every decade. (BC govt)
What’s the risk of a big one’ in BC?
The Cascadia fault line, where the ocean dives under the North American continent, will, one day, cause a very big quake: it is possible this could be above magnitude 9. This, like the Japanese quake of 2011, could also trigger a tsunami. A “really big one” would be devastating: FEMA projects that nearly 13,000 people would die (mainly in the US, because of the higher population density and larger area affected). The odds of a big Cascadia earthquake happening in the next 50 years are roughly 37%; the odds of it being magnitude 9ish are about 10%. Those are rough numbers. (New Yorker)
Researchers have found clues of 13 big quakes in the Cascadia zone over the past 6,000 years, at intervals of 200-800 years apart. The last one was 300 years ago. So another one could happen very soon, or not for hundreds of years. (NRCAN)
Do small quakes relieve the pressure on a fault system and make big quakes less likely?
Not in a significant or practical way. For some kinds of faults, it is true that small slides take some of the pressure off. But, as the BC government puts it, you would need a million magnitude 2 earthquakes every day for 500 years to release enough energy to prevent a magnitude 9 earthquake. (NRCAN)
Can you predict a quake?
No. No one knows how to predict earthquakes. Sometimes people use the word “predict” loosely and claim they have good results, but this is really for earthquake “forecasting” (think about the difference between a weather prediction – it will rain heavily at 3pm – versus a forecast – there’s a 90% chance of rain this month). So, for example, researchers recently claimed good results using artificial intelligence to forecast the highest magnitude quake likely over the next 30 day period in Los Angeles (a very well studied and measured area) (Nature paper). This sounds exciting but it’s of limited practical use: if I told you there was a 90% chance of a magnitude 6 earthquake in your town in the next 30 days, what would or could you do about it? You could make sure you had a plan and supplies, but you should do that anyway. False alarms and unnecessary evacuations can be costly and harmful to health.
Can you get early warning of an earthquake?
Yes! The first whoomp of an earthquake (called a pressure wave) travels faster than the shaking part (called an ‘s’ wave). Detect the first bit, especially near the epicentre, and you can send out a warming faster than the shaking, giving people up to tens of seconds of forewarning. With this time you could: stop trains, close tunnels and bridges, pause surgeries, and dive under a table or your school desks. This is amazing, and they do it all the time in Japan.
BC’s early warning system got up and running on 29 August 2024! This is great. Before this, we had one system specifically to close the George Massey tunnel (which goes under the Fraser River) when a quake was detected. Now everyone can get warnings (Earthquake early warning blog). I think the only time it has been activated before yesterday was in September, for a magnitude 6.5 quake south of Haida Gwaii. It’s not clear to me how many seismometers this runs on, or where they are – the more the better. Oceans Network Canada has some detectors way out in the middle of the ocean, and is part of this effort (great ONC video here).
What else can or should BC do?
Some public buildings in BC aren’t expected to fare well in a quake – including some elementary schools. This is appalling and should be fixed. My friend and happiness researcher Elizabeth Dunn has helped to try to promote this issue in the past (UBC).
The earthquake early warning system is awesome and should be expanded, including by hooking it up to automated systems for closing bridges/tunnels/etc.
What can or should you do?
You should be prepared for any emergency, whether it’s a quake, volcanic eruption, flood, wildfire, zombie apocalypse or US incursion. Have a plan for meeting up with loved ones if you get separated, have basic survival supplies on hand at home, and have a go bag packed in case of evacuation.
