Next was a fantastic talk by Lily Hu on the problems with variables in algorithms and social science more broadly at the #Oxford. This is a deep dive into the impossibility of defining a "neutral" variable or algorithm, and how one has to think very carefully about how to define and analyze variables in systems, and ensure that one communicates those decisions transparently. Highly recommend https://www.youtube.com/watch?v=7FqCSoe04rM (3/8) #AI
Next was an interesting talk by Kacper Sokol introducing a framework for explainable #AI (XAI) at the #UNSW AI Institute. This is a good grand tour of the space if you are unfamiliar with XAI https://www.youtube.com/watch?v=P-LkoI3VpLU (5/8) #XAI
Watched another interview on MSNBC about "human extinction" by "AI", this time with Yoshua Bengio, and am wondering why all those "godfathers" of "#AI" go on TV when they have nothing to say, other than express their "fears"?
Why not do some serious work instead, with data and theory, get results and publish some serious papers about it? Some of their colleagues already do that, building foundations for a legal environment where such technologies can operate safely.
Somebody is making sense about the real dangers of #AI! I completely agree -- I'm not worried about electronic overlords any time soon, but I am very worried about new human overlords (hiding behind the supposed threat of electronic overlords.)
And I would add I'm also worried about clueless people applying hallucinating tech to important things -- sooner or later somebody is going to hook up GPT4 to, well, why not a nuclear power plant. And then we're in for a ride.
«««««
In all affairs — love, religion, politics or business — it’s a healthy idea, now and then, to hang a question mark on the things you have long taken for granted.
»»»»»
As I was unable to find an actual source for the quote through my own searches, I thought I would let the #Google#Bard#AI try.
It found a source. I asked where in the source it was. It gave me a big quote in context. And the big quote (and the smaller quote) are demonstrably not in the book it quoted.
This piece made me think of:
💡 The highly influential AI doomers are all elites.
🤔 Why is it that the elites are so bothered by existential?
🔥 Because anything less catastrophic won't affect them.
📍 AI harms that affect the majority of us don't even touch the elites.
🎯 Thus, the fearmongering comes from a selfish place.
...since starting @ClimateMigration yesterday, the vast majority of Followers are from outside the US. Understand this is not a fun subject, but the disparity in attention to this topic is pretty striking.
@koteisaev Trying to figure out the difference between a curated subject account such as @ClimateMigration and the new hot topic, "controlling your algorithms", and it seems that it is just that some sort of AI replaces a human curator in the latter. Not sure that is really an improvement.
Alright, there is much #buzz over #AI in #edtech. While I embrace what #LLMs offer, why are so many people going mad over the topic? I mean, I'm not a grandmaster in the #education field (I've been doing it for 15 yrs "only") & am pro #innovation, but what's wrong with using less tech?
Tech should be integrated to a degree & definitely taught about, but beyond that it will mostly get in the way. Effective #teaching has less to do with tech & all to do with #methodology.
Funny how when blue collar workers complained about machines taking their jobs we called them Luddites and suggested they retrain in some other profession, but now that AI looks like it might take white collar jobs we're demanding parliamentary and congressional hearings. Funny that.
I tried that new #AI thingy in #Photoshop where you can expand the edges of an image, like people having been doing with memes and album covers. Absolutely mind-blowing how good it is. #design
« So when Bard was first made available to the public, AI researcher Kate Crawford was quick to ask Bard itself where its dataset came from. The answer caught her attention: Bard said one of its data sources was Gmail. »
De Autoriteit Financiële Markten (AFM) maakt zich zorgen over kunstmatige intelligentie. In een woensdag gepubliceerd rapport waarschuwt de toezichthouder dat artificiële intelligentie (#AI) door de huidige snelle ontwikkeling een gevaar vormt voor de financiële stabiliteit.
EDIT: the word "hallucination" is probably out of place and misleading here, as @apodoxus argues in this thread. The anthropomorphic connotations of the word may actually create more misconceptions than help eliminate.
@apodoxus
Indeed I have. Almost everywhere artificial intelligence is mentioned in public, with an exception of sincere technical talk. I read this paper https://arxiv.org/pdf/2305.09800 which struck a chord with me and I recommend to anyone interested in #AI. I have encountered all issues mentioned in the paper while working with my Little Assistant, a mostly symbolic agent for gathering numeric data and other information.