ML + AI advances 2016

hi all. machine learning + artificial intelligence are evolving very rapidly these days. very hard to keep up but heres a major effort to pound the latest headlines into a coherent summary. the media has picked up on the major horserace. a lot is being cast as corporations versus each other, mainly google[f], facebook[g], IBM, Apple, etc… with literally billions of dollars being thrown at it now, at this rate it looks like there is not a lot of question at this moment who is going to have the most cutting edge ML+AI in the near future, and hint, it isnt gonna be universities+academia. its being pitched as the next disruptive technology to hit silicon valley.

have some major qualms about this. there are only a few disruptive technologies per generation. ML+AI does seem to fit in this category but its much more inchoate than eg say web browsers, internet, email, or cell phones. even with good ML+AI its not so clear how to productize/ monetize it, and in a way Google is one of the only companies that has figured out how to do so, and its all based on a few basic concepts: relevant search and matching/ optimizing advertising buyers with ad space wrt ad auctions. and in a way, the relevant search algorithm only indirectly contributes to their bottom line. hearing about IBM putting millions of dollars into Watson research makes me wonder if theres definitely a light at the end of that tunnel. and whether its an oncoming train. etc

advances in this area ML+AI are so diverse and are hard to classify, but key ones tend to be in hardware[b] with talk of specialized systems, natural language processing[c], prediction[d], and biology eg breakthroughs predicted or announced in drug discovery.[e] actually the headlines on drug design are sensational and one wonders if it will be one of the major “killer apps” of ML+AI in the future. there are billions of dollars yearly worldwide riding on drug development, & there is increasing focus on “customization” also in cancer research, and it seems like a great, nearly natural match/ alignment/ niche for ML+AI.

there is a huge amount of fear+warnings+progress reports and its not easy to separate all those out.[i] many predictions for the future including worries about it affecting job rates/ employment.[j] and then theres serious to even grave/ alarmist concerns about its impacts.[k] 

lots of edu materials.[l] profiles on leading thinkers/ visionaries/ researchers.[m] advances in understanding the brain tying in with ML+AI technology.[n]

but also talk of inherent limitations.[o]

there are some new contests.[p]

Microsoft had a PR debacle with chatbot Tay picking up bad habits from its ill-intentioned/ nefarious users but it does show how far the field has advanced.[q] then there were still many headlines on the milestone alphago win and another match with another player is being arranged maybe this year.[r]

a major area of innovation is combining ML+AI+ brain imaging technologies to analyze/ understand human thoughts/ emotions in a way (eg high precision, near realtime) that is unprecedented and increasingly revolutionary.[s]

there are many great new books on AI,[t] read a few myself recently in a several months-long binge, was thinking of reviewing a few here but that would take a lot more effort and will wait to do that until my hits on this subject boil over, “lol”. for now, after decades of “not overly impressive/ marginal/ striking” advances in the area, am enjoying and savoring this remarkable moment in the zeitgeist where hype and progress are not entirely disconnected.

& do want to take the moment to highly recommend the very rosy/ optimistic/ pragmatic/ engineering-oriented book by Domingos; its very unusual and serves as a very strong counterpoint/ juxtaposition to the alarm in eg Bostroms book. Domingos has virtually nothing to say about the same topics Bostrom raises and maybe thats not a bug but a feature. (its also striking to compare Bostroms writing essentially/ centering on philosophy to decades old musings of legendary AI nemeses eg Searle on supposedly the same subject; they are nearly diametrically opposed because Bostrom is interested in/ focused restraining superintelligence, and Searle asserts emphatically AI will never happen/ be achievable to begin with… but philosophy can be like that sometimes…)



a. advance
b. hardware
c. NLP
d. prediction
e. bio
f. google
g. facebook
h. corp
i. fear+warning+progress
j. future
k. bad
l. edu
m. profile
n. brain
o. limit
p. contest
q. ms tay
r. alphago
s. thoughts
t. book

1 thought on “ML + AI advances 2016

  1. Pingback: d127: ML + AI advances 2016 – AI

Leave a Reply

Fill in your details below or click an icon to log in: Logo

You are commenting using your account. Log Out /  Change )

Twitter picture

You are commenting using your Twitter account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )

Connecting to %s