Mark Crowley

@[email protected]
51 Followers
18 Following
32 Posts

Associate Professor at the University of Waterloo
Electrical and Computer Engineering
#UWaterloo

Research on
#ArtificialIntelligence #MachineLearning #ReinforcementLearning
#AI #ML #AIML
#AI4Science #causalinference
#manifoldlearning #dimensionalityreduction
#compsust
#forestwildfire

Computing takes thinking.
Thinking is computing.
Intelligence is the fixed point of this loop.
AI is a convergence.

Joined Mastodon: Apr 2017
Main account: @compthink

website:https://markcrowley.ca
main (for now):@compthink
ORC-ID:https://orcid.org/0000-0003-3921-4762
on birdsite:@compthink
Show threadMark CrowleyJul 7, 2023
@dingemansemark gasp!
Mark CrowleyJul 6, 2023
kv Jul 6, 2023

Please remember to donate, especially to smaller and more specialized instances if you're a member! You're not a product on the fediverse unlike #threads et al so it's a cost to owner/operator's to run!

@maker https://handmade.social/@maker/110668851071401998

handmade.social admin :maker: (@[email protected])

I have to admit that I am very discouraged by the fact that it was only non-members who contributed to the expenses of keeping handmade.social running. In the wake of my announcement that handmade.social would remain operational, member financial support has been non-existent. All it takes is each member pledging $1/mo to keep handmade.social going. Otherwise, we will be shutting down in 12 months instead of 3. If you value this instance, please donate > https://buymeacoffee.com/handmade.social

handmade.social
Show threadMark CrowleyJul 6, 2023
@ondra @Yoav a mesh?
Show threadMark CrowleyJul 6, 2023
@bkeegan I've never even heard of Horizon worlds, was that Meta's VR thingy?
Show threadMark CrowleyJul 6, 2023
(posting this on "all three" places to compare conversations, don't worry, I don't make a habit of crossposting)
Mark CrowleyJul 6, 2023

So very many questions really:
- define "smarter than us"
- define "human intent" in a completely general way covering all possible situations
- what if the eventual #SuperAI is "smarter" than your #Alignment Researcher?
- does alignment mean the system does exactly whatever the human intent is?

What is the intent is bad? Some humans have bad intents. In fact, all humans have bad intents sometimes, either intentionally or not.

In the end, humans need to be responsible for things we make.

Show threadMark CrowleyJul 4, 2023
@harriorrihar I can't argue with that.
Show threadMark CrowleyJul 4, 2023
@harriorrihar indeed, first priority is to ensure they don't somehow control this ecosystem by joining it. but second would be peacefully maintaining interopability so that the Fediverse is connected to a larger network.
Mark CrowleyJul 4, 2023
Bernard TyersJul 4, 2023

My best wishes to Facebook with their new ActivityPub based social media twitter thing.

It will be amazing and they will make the world a better place. Of course they will.

I'm sure lots of people will try it out. Don't hate on them for doing that.

Just remind them of the interesting permissions and data gathering the "Threads" app, and therefore Facebook does, in comparison to the Mastodon app.

Which would you choose?

#activitypub #mastodon #threads #facebook

Mark CrowleyJul 4, 2023
Do Mastodon instances usually have agreements with other instances? I thought the idea was they do what they do, and if you loke it you can block their instance. Same for #Threads