haldujai

47 ms·

481.
by haldujai 3y ago
> 3. We have not built superhuman AGI. But superhuman AGI is probably possible. "Dumb" AGI is yet to be proven theoretically possible let alone superhuman.
482.
by haldujai 3y ago
"not taking these things seriously" is very different from regulating something in it's infancy because of fear of the unknown. > And might I point out that a form of bioengineering may have caused the latest global pandem
483.
by haldujai 3y ago
> I can appreciate that. I haven't been listening to him for that long; having no idea what situation you're discussing, I'll take it at face value. https://m.youtube.com/watch?v=2HMPRXstSvQ We are current
484.
by haldujai 3y ago
I believe the burden of proof is on the one making the claim. Misinformation has been a problem for all of human history, in its current form dating back to at least the Spanish civil war. All the existing version of AI does is replace trol
485.
by haldujai 3y ago
Hinton has been exaggerating the capabilities of AI for nearly a decade. As one example, if we listened to him in 2016/17 we would have an even worse radiologist shortage than we do right now. > The point is not that any one particu
486.
by haldujai 3y ago
You mean an industry with so much inertia and hesitancy they still run mission critical systems on tech from the early 2000s? You think they’re dumb enough to risk throwing barely functional AI in there on the chance it’ll save a few bucks
487.
by haldujai 3y ago
There is zero chance congress adds regulatory hurdles to the NSA to stop them from building PRISM-LLM in the name of protecting the homeland. The only thing that can be regulated is private citizens and corporations.
488.
by haldujai 3y ago
What is so impressive about ChatGPT that it poses an existential threat? Many experts are highly critical (e.g. LeCun) and/or believe LLMs are nothing more than stochastic parrots. Anyone who was working with transformers could have se
489.
by haldujai 3y ago
Borrowing your analogy: Except right now AI isn’t an ICBM or sarin gas, it’s a slingshot, and we don’t even know that current approaches will get us there. AI risk is a valid discussion but regulating AI in its infancy seems misguided at be
490.
by haldujai 3y ago
While true the ones doing media tours and speaking the most vocally in May 2023 are the LLM crowd. I don’t think it’s a mischaracterization to say OpenAI has sparked public debate on this topic.
491.
by haldujai 3y ago
I think you’re misunderstanding my point. I never said “they should have disclosed a conflict they did not disclose.” Disclosures are absent from this initiative, some signatories have self-identified their affiliation by their own voliti
492.
by haldujai 3y ago
So the most “notable” AI scientists on this list have clear corporate conflicts. Some are more subtle: > Geoffrey Hinton: Emeritus Professor of Computer Science, University of Toronto, He’s affiliated with Vector (as well as some of the
493.
by haldujai 3y ago
I’m not painting anything, if a disclosure is needed to present a poster at a conference it’s reasonable to want one when calling for regulation. Note my comments are non-accusatory and only call for more transparency.
494.
by haldujai 3y ago
Yes, as I said “many” have obvious conflicts from listed affiliations so it would be nice to have a positive/negative disclosure from the rest.
495.
by haldujai 3y ago
> will lead to the project approvals, grant money, etc. In other words, a potential conflict of interest for someone seeking tenure?
496.
by haldujai 3y ago
Huh, you don’t have to do any research. Go to: https://www.safe.ai/statement-on-ai-risk#signatories and uncheck notable figures. Several of the names at the top list a corporate affiliation. If you want me to pick specific
497.
by haldujai 3y ago
I used LLM because the people shouting the loudest come from a LLM company which claimed their newest language model can be used to create bioweapons in their whitepaper. Semantics aside the recent interest in AI risk was clearly stimulated
498.
by haldujai 3y ago
Why do you have to inoculate yourself from it to create havoc? Your analogy of “nuclear war” also has no vaccine. AI autonomy is a hypothetical existential risk, especially in the short term. There are many non-hypothetical existential ri
499.
by haldujai 3y ago
It’s unequivocal that academics may have conflicts (in general), that’s why disclosures are required for publications. I’m not uncovering anything, several of the academic signatories list affiliations with OpenAI, Google, Anthropic, Stabil
500.
by haldujai 3y ago
I mean a small group of malicious humans can already bioengineer a deadly virus with CRISPR and open source tech without AI. This is hardly the first time in history a new technological advancement may be used for nefarious purposes. It’s a
501.
by haldujai 3y ago
> There are quite a few researchers from Stanford, UC Berkeley, MIT, Carnegie Mellon, Oxford, Cambirdge, Imperial College, Edinburg, Tsinghua, etc who signed as well. I know the Stanford researchers the most and the “biggest names” in LL
502.
by haldujai 3y ago
Having their names on something so public is definitely an incentive for prestige and academic promotion. Shilling for OpenAI & co is also not a bad way to get funding support. I’m not accusing any non-affiliated academic listed of doin
503.
by haldujai 3y ago
Subtract OpenAI, Google, StabilityAI and Anthropic affiliated researchers (who have a lot to gain) and not many academic signatories are left. Notably missing representation from the Stanford NLP (edit: I missed that Diyi Yang is a signator
504.
by haldujai 3y ago
> BERT-like encoder models actually do help here, although unfortunately the dense representation is not as nice and clean as a class, it will at least map to something similar. Completely agree with this. > If the hospital provides e
505.
by haldujai 3y ago
There are several corporations, some of which have raised > 100M, tackling this problem. AIdoc being one of the more notable ones.
506.
by haldujai 3y ago
I agree it's not impenetrable, that's why I'm working on this problem. What I disagree with is the "this is trivial" statements. > Unfortunately there's a lot of stupidity going around right now in thinking
507.
by haldujai 3y ago
You're missing the point, it takes me < 5 seconds to clear the adrenals. The example is intended to illustrate that there is no feature to extract that would make a model BETTER than a human for the things that we care about (rare a
508.
by haldujai 3y ago
Exactly this! I chose radiology given my background thinking I could "easily build radiology AI systems" and help our struggling system. Then I became a radiologist and quickly discovered how hard this is. Something "as simpl
509.
by haldujai 3y ago
Speaking to radiology, ~90% of imaging studies are normal so I can sign my default template without looking at the images and be correct the vast majority of the time.
510.
by haldujai 3y ago
I've been involved in a federated learning radiology project, the biggest issue is image technique and labelling. Different centers practice very differently, with different imaging protocols, disease prevalences, and labelling/re

More ›