comprehensive answer to "Is it OK to use AI?" by @gwenvarley.bsky.social
an excellent one for students and those mired in the terminological and ethical disarray to start to think more clearly
California boy dies of SSPE (measles complication) after being infected at age 7 months, when he was too young to be vaccinated.
Not a surprise. Code LLMs have lower temperature params. The first result would be close to the high statistical score. Any revision is going to include lower scoring results but may never get to the point of settling on the result needed if there aren't enough examples in training.
He should book a cabin on one and make the voyage.
Then make sure the patches are up to date.
The Guardian has seen the pictures of his tattoos, right?
Absolutely this.
I would like to add they need to realize any product that must be force fed to people who then have to have access to expert knowledge and take heroic action just to limit how deep it gets into their personal space is meant to be a tool of surveillance and oppression.
I don't have experience with voice to text bc reasons, but I know that word prediction, spelling and grammar prediction have gotten a lot worse. This can be tracked to the design of transformer architecture which, in most cases, favours interesting output over just correcting the input.
The shitty part of it all is that before modern LLMs, we had amazing technology that provided accommodations for so many people.
I think of things like: customized words/phrase prediction, autocorrect, suggestions, voice to text and vice versa, etc.
And now NOTHING is not enmeshed with "a.i."
"The United States is the only place we surveyed where more adults (ages 18 and older) describe the morality and ethics of others living in the country as bad (53%) than as good (47%)."
Never base financial decision on your own marketing BS.
Two white nationalists I exposed in a recent article showed up to try to intimidate me in person last night.
This is a press freedom issue.
Here’s what happened:
Tell me I can order these somewhere.
AI is not inevitable. Nothing in human societies is inevitable because we design them. Healthcare can be free for the public. Books can be bought instead of bombs. Universities can be free for students, and they can even receive a stipend to live off. Don't let companies dictate the future.
A minion of an Orwellian regime quoting Orwell would be peak irony. Misquoting him is a little too on MAGA brand.
300 characters are never enough to express a complex idea. I never thought for a moment it wasn't be legit.
Doesn't sound fake to me at all. I've know people like that including one who, with his team, stepped out of an airplane at 1500 feet in the high arctic in the cold, dark, and strong winds of October, parachuting onto a crash site to begin the work of saving 13 survivors including my good friend.
Corollary to quoted IBM training: "A computer can never be convicted of war crimes, therefore a computer must never make any combat decisions."
One more reason I'm glad I've never used Grammarly.
It’s not that there are zero use cases for AI.
It’s that there are very, very few that justify the enormous expenditure of resources and rapacious exploitation of labor.
In the 1990's I worked on a successful 'AI' project. Not near the size of LLMs today but it was very useful. The biggest part of my job was curating the data that would be fed back in for ongoing training. Those enthralled by the tech were uncritical of any result produced, sound familiar?
"The IT Crowd" was originally supposed to be a documentary.
*training data*
You are letting anthropomorphism do too much work for you. The 'I' comes from the fact that it is the most common syntactic element found in the training day that most closely resembles the exchange of tokens happening. That is really the best you can hope for from any LLM in any circumstance.
If you use this in Rome to you remain under the control of your father, sine manu, or are you now under the control of Zuck, cum manu. The latter is kind of scary.
en.wikipedia.org/wiki/Manus_m...
@olivia.science Never (knowingly*) used it. In this house we only use AI I create myself on a replica of the MIT AI Lab DEC System 10 simulated on a Raspberry Pi.
*It is difficult to be sure as like Franks Red Hot they put that sh*t in everything. Skipped a meeting last night that used copilot. 😅
THEY MURDERED 60 KIDS IN AN ELEMENTARY SCHOOL YOU FREAK
This is a mistake. The action the US has taken in Venezuela, Cuba, and now Iran will further destabilize international relations and encourage additional conflict around the world.
