ericpanzer.bsky.social
Arguing since birth. Urban planning, politics, weather, Drag Race, and photography are all fair game. Opinions are strictly my own—unless others agree. I assume that strings of emojis in people's bios are their washing instructions
717 posts
623 followers
396 following
Regular Contributor
Active Commenter
comment in response to
post
A coinflip has no priors, so to speak, all of the things I listed above, including LLMS, effectively do
Demonstrating this is very simple. Ask an LLM why the sky is blue. It is much more likely to give give/lead you to an accurate answer than by playing 20 questions with a Magic Eightball
comment in response to
post
But your claim was that it has the same relationship to truth as a Magic Eightball, it does not because an Eightball's answers have no relationship whatsoever to the sum of human knowledge
LLMs *do* have at least some relationship to this knowledge, even if that relationship is not 100% reliable
comment in response to
post
The question is more about institutions than it is about technology, honestly. You can have a lot of technology or next to no technology, and the question will always been who is empowered to manage information and/or be considered an arbiter of truth
I agree, LLMs are not a replacement for this
comment in response to
post
The thing is, there is no situation where one is seeking information that does not call for some level of human judgement
According to most media in 2002/2003, Iraq had a program to create weapons of mass destruction. There is no such thing as a trustless information environment for human beings
comment in response to
post
The probability of each of these things providing accurate information rests on a variety of factors, including the intentionality of how the information in them is curated, analyzed, presented, etc. For all the things I listed above, as well as LLMs, the result is not based purely on a coin flip
comment in response to
post
I mean the question of whether a completely hallucination free model is possible is outside the scope of this discussion. But there's no such thing as hallucination free information period because humans also hallucinate (flat earth, anti-vax, etc.)
The question is the relative accuracy of methods
comment in response to
post
A search engine, a website, and a book do not "understand" your questions, but they are all still at least capable of providing/connecting you to accurate information
comment in response to
post
I know you don't like LLMs or AI in general, and that's fine. There are reasonable criticisms and certainly negative consequences to address. But when you make false claims like this, you undermine efforts to promote *healthy* skepticism and measured approaches
comment in response to
post
Saying it's fundamentally a Magic Eightball is itself not a remotely true statement and it's ironic you would promote such misinformation as you worry about LLM's accuracy
The internet is not 100% accurate, but it is still an incredibly useful tool for research and finding information
comment in response to
post
I know this site is inclined to yeet me into the sun for saying this, but...
Even though LLMs should absolutely not be relied upon as an ultimate source to determine what is true, they can be made to offer sufficiently accurate answers to be useful in many (but certainly not all situations)
comment in response to
post
Oh I'm not saying I don't get angry. I'm just saying the anger feeds me. And it's really only some small sense of propriety that generally keeps me from verbally destroying people
comment in response to
post
When will State resources be deployed to prevent these extrajudicial kidnappings by unidentified, masked secret police?
Whatever we think people should have done to resist the SS and Gestapo, that's what we should be doing *right now*
comment in response to
post
Most mainstream press, including the SF Chronicle, have basically turned into glorified stenographers (or worse), with little to no context or investigation
I was able to pull up suspicious info about BSF with <15 minutes of web searches
@sfchronicle.com owes @mrsleather.bsky.social better
comment in response to
post
I'm going to put in another plug for Make Shit Cheap Again
comment in response to
post
He's more than a phony, he's a Nazi collaborator
comment in response to
post
You are a Nazi collaborator
comment in response to
post
comment in response to
post
People would do well to remember, though, that the Night of the Long Knives (which took place in 1934) did not result in the fall of the Nazi regime, which would still go on perpetrate the Holocaust and start WWII over the following 10 years
comment in response to
post
A lot queer folks, most notably Millennial and younger, seem to think that the implosion of Rainbow Capitalism will lead to Progress Pride Flag Socialism, when the most likely outcome remains Khaki and Navy Fascism
comment in response to
post
Fox News by EOD: A zebra is a horse in drag, and should therefore not be allowed to read to children
Tackling the tough issues
comment in response to
post
Or maybe don't sit by with popcorn and instead vote against all Trump's nominees and get out there like AOC and Sanders making the case that these are reckless people hurting America
comment in response to
post
Letting it *run*, but saving money and sticking it to big companies trying to profit off bad air by selling overpriced devices is also fun!
comment in response to
post
Important PSA: you don't need a fancy multi-hundred dollar air purifier. Slapping a MERV 13 20x20 furnace filter on the side of a box fan and letting it fun in a closed room is roughly as effective
If you want to be fancy, make a cube out of filters
cleanaircrew.org/box-fan-filt...
comment in response to
post
@therealjeanwalsh.bsky.social
comment in response to
post
Also, one of my biggest criticisms of Harris's campaign message is that it was too focused on things like tax credits for starting a small business. That's lovely, but that's not what most people think they need to get by and get ahead
comment in response to
post
There is more to focusing on kitchen table issues than talking about them during the course of a Prez campaign
In any number of ways Dems are better on these issues, but blue states and coastal cities have let cost of living (most notably housing) spiral out of control
We must focus on results
comment in response to
post
Don't be silly, we can't be delivering on things like housing, we have neighborhood character and landmarks to protect!
And how will low income and marginalized groups know just how much we're listening to them if every policy, plan and program doesn't have 18 months of community engagement?
comment in response to
post
I see everyone who doesn't plan out their categories before making their guesses as chaotic neutral, but maybe y'all see me as lawful evil?
comment in response to
post
Like GMO before it, AI appears to have entered the catechism of technologies the left must hate because apparently we can't separate a technology from its worst proponents/users
comment in response to
post
As a dyed-in-the-wool bleeding heart pinko liberal, I find it so strange that the default progressive tribal position has become so negatively polarized against AI that the same people who are *very concerned* about climate change are now defending desert-grown alfalfa for *beef and dairy exports*
comment in response to
post
Conservatives must hate vaccines and renewable energy
Liberals must hate GMOs and AI
These are the beliefs we were assigned by JK Rowling's Cursed Tribalism Sorting Hat when we arrived at The Bad Place
comment in response to
post
The "some" is doing a *lot* of work in that sentence.
comment in response to
post
Not everyone has my Jewish super power of feeding off criticism like Ungoliant sucked up the light of the Two Trees
comment in response to
post
Interestingly, the hissing culture seems to have faded in favor of murmuring or outright yelling
I miss the hiss. It was cute.
comment in response to
post
And for those who are going to say "but the environment," I guarantee that you the marginal impact of using AI to read names is far, far less than that of manufacturing the robes, growing the flowers people give each other, or having people drive to the graduation venue.
comment in response to
post
Pointless but ultimately harmless? Most of us probably can't name or conjure the face of the person who read the names at our graduations. Anyone for whom that's the case probably shouldn't be terribly exercised about this.
comment in response to
post
Composed of elected officials from jurisdictions across Alameda County, it perfectly demonstrates the results of selfish parochialism
Under the guise of promoting green building, the board voted to preserve localities' ability to subject new housing to ever-moving targets of development standards
comment in response to
post
AI would be absolutely amazing at analyzing meteorological data, too bad we've cut the NWS to the bone and have dramatically curtailed collecting the very data the AI would need
comment in response to
post
About an hour ago I was out at a bar with my partner and friends and felt joy and laughed any number of times
On the way home, the driver has had Alice on and I got emotional thinking about how my family would visit the Bay when I was a kid and Alice 97.3 seemed *so cool* to me