(Audio Version) Decoding Academia: Matt Edition *UNEDITED*
Added 2021-11-25 10:23:11 +0000 UTCVery similar to the video version... here is Matt's heroic tale of his academic journey.
Gather the kids around the fire and nestle in for a tale of daring, statistics, and convulation.
Enjoy!
Comments
I don't know if we could do a normal episode on AI but maybe with the right tech guru. We want to do Jaron Lanier but he isn't exactly the right fit. I do think the topics you raise Matt probably has a lot of opinions on though. Maybe a bonus episode?
Christopher Kavanagh
2021-11-27 06:35:13 +0000 UTCOnly by those who follow conventions.
Christopher Kavanagh
2021-11-27 06:26:48 +0000 UTCMatt’s use of language is very interesting in that he naturally uses older pronunciations of common words eg. “convulation” and “somethink”. Would love to hear both of them (especially Matt given his penchant for sci-fi) use the word “grok” more often.
elcid
2021-11-27 01:36:29 +0000 UTCOr "convolution", as it's also spelled.
Duncan
2021-11-26 19:19:32 +0000 UTCSo glad you guys decided to do these back-grounders! Matt's path was fascinating and I have some overlap that I also find quite interesting - not in stairs building, though. I started on a PhD in Neuroscience in the 90's and left it, before I finished, for software design in the early 2000's. I had a similar reaction to the fMRI stuff ("correlations with blobs of activity"), did a bit of developmental biology with Zebrafish and came back to the psychology department to do brain surgery and behavioral studies on rats and their amygdalae before starting my career in software design. Two of my professors in the psych dept may be known to Matt -- their work was also focussed on addiction and I really enjoyed the seminars they led: Kent Berridge and Terry Robinson (academia is so siloed and varied, one never knows what overlaps what, but I have seen them both only grow in reputation since I studied with them so there's a chance that Matt knows of them). Two topics you hit upon in this discussion make me really hope you could turn them into future DtG episodes: 1. A revisit to Taleb's claims regarding statistics and life with a deeper dive into the statistics based on Matt's capacities in the area. The main problem I can see here is that it might require reading several of his books to nail down what his arguments are and I don't know if that sort of time investment is in the cards/worth your time. Taleb is a bit maddening as he ranges across so much territory and seems tantalizingly close to having something valuable to add about types of risk and reward distributions and how it makes sense to live, given these -- but his wide range is also a bit obscurantist -- he seldom sits still long enough or makes clear enough claims for anyone to solidly evaluate/criticize what he has to say. He repels critics by jumping from topic to topic and no one sees themselves as expert in all these areas as well as by practicing his over the top verbal punishment of anyone who dares question him such that serious critics probably seldom feel it is worth their time. Anyway, it would be fun to hear from someone who feels comfortable with the math and stats try to decipher his claims and separate the possible wheat from the chaff. 2. This one is more "out there", but I think could be VERY cool: address AI as a guru to be decoded. Surely one can imagine a personified AI -- especially in the boogey-man stories that tend to get the most attention in the popular press -- the fear is essentially AI agents becoming our rulers. I know its a stretch, but there is at least one story that really needs to be told and Matt appears to be in a great position to tell it -- I "left" neuroscience in 2003, just as Matt was getting started. I was still involved in tech though and worked with Google though not at all deep in the tech sector -- much more in the end-user/design side of things. The strange thing is, and I wonder how many people had this same experience, when I was deep into Psychology and Neuroscience the main narrative around AI was that it was a failed pipe-dream -- and most likely for good reasons -- the attempts to make machine analogs of a system that -- as Matt pointed out, we basically understood at the "blob of activity" level seemed silly. The few systems that people tended to tout were only analogous to brains in very simplistic ways. It took me a very long time to accept that this latest wave of AI was really a breakthrough and really potentially useful because I thought I knew better from being very aware of past attempts. I'm still skeptical -- just what is machine learning and does it have anything to do with the way brains work? does it really have the potential to power evil entities who take control? or Scarlett Johansson-like bots with whom we can have magical relationships? OR is it just a form of adaptive computation that can have great impact but doesn't really fit with boogey-man/scifi AI, reflect anything about how our brains work or deserve the name "intelligence"? Is it too much of a stretch? Could you imagine an episode on this topic?
Tom Allison
2021-11-26 14:11:42 +0000 UTC“convulation”?
Tom Allison
2021-11-25 11:46:38 +0000 UTC
