I recently discovered that I’m the only one in the #birder community both taking a flâneur approach to the whole thing, and, importantly, using pataphysics as both a method of inquiry and as a taxonomy.
#philsci #philosophy #philosophie
1. One is an artifact, the other one is not: Diff. Epistemic norms/practices
2. One is “Essentially opaque” (Humphreys) in an agent neutral and/or agent independent way (me), the other seems to be only generally opaque.
3. Trust in scientific results is separate from trusting scientific practices.
I use to buy this and counter that artifacts shouldn’t be as opaque as we are. Later I realized its false premise: we actually don’t trust opaque humans. We provide education in best practices, licenses, and transparent methodology that we can rely on or inspect for scientist, doctors, judges, etc.
To continue comparing humans to artifacts is weird. It’s like comparing hummingbirds to jet fighters and then pointing out their similarities to dismiss arguments.
Completely. And even different from other computational methods. The notion of “strange error”, which you mention and some of us have been writing about, is also very central to the kind of opacity At play in AI.
Arguments were good. People hadn’t yet thought of instrumentation that way in philosophy of science. Computational methods weren’t really understood as particularly interesting until recently in #philsci except for a few exceptions. AI has made them inescapable.
plato.stanford.edu/entries/simu...
Even further, the kind of opacity in computational artifacts is distinct in both its nature and its extent. It may be resistant to reasoning strategies meant to survey for epistemically relevant aspects: source of error, inferential paths, bs, etc.
philsci-archive.pitt.edu/19384/1/Expl...
In fact we are not. The processes by which we are trained are transparent: education, licensing, practice conventions, methodology, etc. We don’t trust the humans, we trust the process which we can inspect. Furthermore, artifacts are different from natural processes which we have little to do with.
Some of us in philosophy of science have been concerned abt this wrt computational methods for almost three decades (before AI) without anyone really listening. The Epistemic opacity in scientific computer simulations literature started bc of this concern. Glad people are catching up.
#philsci
This is something that some of us in philosophy of science have been concerned abt wrt computational methods for almost three decades (before AI) without anyone really listening until now. The whole epistemic opacity (Humphreys, 2004) in scientific computer simulations started bc of this concern.
🫨Such an awesome start to our microworkshop on #AI and Close Personal Relationships. Amazed at the beauty of this model of academic interactions. Beats any conference!
The perfect set of interlocutors on AI and friendships, love, etc. + an intimate setting for work in progress.
#philtech #aiethics
2 days 3 talks! #AI and Loneliness, Artifactual Epistemology and AI, and Civilizational Risks and AI. First one to the greater community at Iowa State University, second to their philosophy department, third to a life learning community of retired academics in Oregon.
#philtech #philsci #AIethics
I was thinking about this the other day and I had absolutely no interest in being 20 again. I had a blast, had a band, hitchhiked, etc. but it depends on what you mean by “freedom”. I am a lot freer to do what I want with life now, I’m also exactly where I wanted to be then (minus a few aches).
Sometimes I know what I’m looking for but don’t know where I’ll find it. Sometimes I don’t know but then I find it. Sometimes I am not looking for anything and something finds me. As the history of science has shown, serendipity is often enriching for research and the researcher.
#philsci #philtech
Automating SOME specific aspects of research/science may be fine. Doing so uncritically or for bad reasons is not. Thinking that something being “better” (faster/cheaper/more accurate, etc.) is by itself reason enough for it to replace something that is not, is a fallacy.
#philtech #philsci
Yes and no. It does lose some of the anthropocentric aspect of the question, but it opens up for genuine understanding of what I call “artifactual epistemology”—the idea that artifacts can carry out Epistemic operations WITHOUT being intelligent. This is still a philosophically rich possibility.
I philosophize about #AI. Nothing that I have written or will ever write will ever use it. Not bc I distrust it or bc I think it couldn’t do it. Just bc, to me, figuring out ways to do some of the things it could do is part of the beauty, privilege, fun and the point of what I do.
#philtech
Turing rejects this in the first part of that paper (they knew each other). He thinks it is silly to rely on the way people use concepts.
Why do people think the Turing test is abt machine intelligence? It’s clearly about OUR ability to discern/tell the difference given ONLY output/behavior concerning discreet tasks. Two completely different things. If you think otherwise, you didn’t read that paper correctly.
#AIethics #philtech
The lawful/unlawful angle wrt Anthropic + DOW debate seems like a red herring: mass surveillance of American citizens could become legal, that doesn’t make it any better/desirable/moral. Autonomous weapons could be legal, that doesn’t mean AI is ready/reliable/safe for them.
#AIethics #philtech
And even quicker… I guess someone got tired of winning the AI race?
#aiethics
Well that escalated quickly…
#AIethics
Our deadline for abstracts for the international association for computing and philosophy is tomorrow! If you are writing on philosophy and algorithms, AI, computers, models, etc., I hope you submit your abstracts!
#philtech #philsci #ai #aiethics #philsky
I’m not sure this is a ‘American liberal’ party line since this could be ascribed to a libertarian as well. Anthropic is heavily enmeshed in US military infrastructure, but they’ve always held the line on these two principles. So, if anything they are being consistent.
"The 45-second film, made around 1897, was the first appearance on film of what might be called a robot"
HELL. YES.
I really enjoy public talks, considering the audience before hand and then their questions at the end is the best exercise to make my ideas clearer. My writing and my lecturing are unavoidably enriched by these events. Here I’m talking about #AI and Loneliness at OSU’s PRAx.
#philtech #aiethics
Please share. The International Association for Computing and Philosophy is one of the best communities to engage with if you are doing any philosophy (epistemology, ethics, metaphysics, etc.) related to computational technologies.
Submit your abstract!
#philsci #philsky #philtech #AIethics
