@max_spero_@s_s_1_9@graykevinb what evals? Your own website states you can’t definitively prove whether post-2022 writing is or is not AI. The fact that you’re accurate in intentionally designed tests does not mean you have any validity in the wild.
@albedoa@habibislop@MaleManlpulator@graykevinb@max_spero_ Sure. You think they realize that this would expose the fundamental flaw in their methodology and that’s why the bounty is only $3K? Not worth it for the average person to go to a library and search for text that fits the bill.
AI interfaces nowadays are still pretty much turn-based.
In my PhD thesis, I thought about what interaction would look like if we could go beyond taking turns.
One example is intra-turn interaction:
Their authority is modern day McCarthyism for white collar workers. “Reasonably reliable” is a disastrous standard relative to the penalty academics face due to anti-AI hysteria that Pangram pushes and capitalizes on.
I don’t have some obligation to help them fix an inherent, obvious flaw in their methodology.
@habibislop@MaleManlpulator@graykevinb@max_spero_ You both don’t understand that the premise is flawed. Anything post-2022, they simply conclude that it cannot be verified whether it’s a false positive or not. Pangram then refers to its pre-2022 ‘authority’ as evidence of its post-2022 accuracy.
It’s a flawed, circular loop.
The other issue is that the only way to disprove them is to pay them lol.
Here’s my general premise, which is verifiably correct. AI writes based off of sophisticated writing principles, of which Garner is on the Mt. Rushmore.
But I’m not going to spend thousands of dollars on Pangram credits to disprove them. Hopefully the public can just come around to common sense. They’re peddling snake oil.
Pangram is a disaster for academia. Its methodology penalizes sophisticated writers at the expense of average writers.
The overwhelming majority of writing that passes as ‘human’ is unsophisticated, human slop.
But AI excels at non-elementary writing, as do sophisticated human writers. So now sophisticated human writers have to dumb down their own writing in order to avoid Pangram’s scarlet letter.
Don’t believe me? Look at @BryanGarnerLaw’s Law Prose blog posts. Anything pre-2022 is marked as human. The overwhelming majority of his post-2022 posts are flagged as AI generated.
What’s more likely?
(A) Garner, who meticulously published musings for decades on how to be a better writer suddenly began outsourcing all of his thoughts to AI; OR
(B) Pangram’s methodology is fundamentally flawed and its credibility rests on pre-2022 writing being hard-weighted in its training protocol?
Pangram is a disaster for academia. Its methodology penalizes sophisticated writers at the expense of average writers.
The overwhelming majority of writing that passes as ‘human’ is unsophisticated, human slop.
But AI excels at non-elementary writing, as do sophisticated human writers. So now sophisticated human writers have to dumb down their own writing in order to avoid Pangram’s scarlet letter.
Don’t believe me? Look at @BryanGarnerLaw’s Law Prose blog posts. Anything pre-2022 is marked as human. The overwhelming majority of his post-2022 posts are flagged as AI generated.
What’s more likely?
(A) Garner, who meticulously published musings for decades on how to be a better writer suddenly began outsourcing all of his thoughts to AI; OR
(B) Pangram’s methodology is fundamentally flawed and its credibility rests on pre-2022 writing being hard-weighted in its training protocol?
@max_spero_@graykevinb You have constructed an illusory authority that can’t be disproved.
“We’re ‘perfect’ on all text written pre-AI, which was our entire training set.
For text written post-AI, we can’t verify accuracy, and neither can you.”
can the pangram fucktards stop acting like $3K is some incredible bounty? If the company was so successful and confident, why not offer an actual prize?
can you imagine growing up from birth through early adolescence under the watchful eye of a ring camera, inside your bedroom
what would that do to a mind
Yes, Jensen Huang slept with a Chinese spy, but never forget that Eric Swalwell also slept with a Chinese spy while sitting on an intelligence committee.
Democrats allowed him to stay on the intelligence committee after he was caught.
@grok please confirm.
Stuff like this is ALWAYS a workflow issue.
So tired of human error going viral as an indictment on LLMs.
FREE TIP: as a last check before placing an order, either have it spawn a subagent to do an ‘oddity check’ (or some other word) for uncommon ordering decisions OR just have it send you the cart and ask you to confirm. Manual confirmation takes 30 seconds and would’ve avoided this.