I can't thank you enough for an overview of AI and large language models. I won't use them and now I know why I will continue to resist using them. I'm glad that I retired as university professor in 2018 before all this took off. I don't know how teachers and professors are coping with this access by students to AI.
I just need to know what happened to the esteemed NYT- they excoriated Biden within an inch of his life and they give this joker a pass on everything WTF
As a computer scientist but not an AI researcher, I humbly suggest the reason AI bots "hallucinate" -- an unfortunate word; "lie" is more appropriate -- is that they are programmed to be good at generating natural-language sentences. These are strings of words that are grammatically correct. However, they do not possess a "model of the world" against which to judge the truth of those sentences. Ask your favorite bot, "Do you have a model of the world against which to judge the truth of your statements?" Humans have such models of the world. (It would be nice if some people didn't have such warped models of the world, but such is life, and such is the state of education. :-))
So, bots sometimes tell the truth, though of course they don't "know" that. The material they were trained on contains predominantly true statements, which means it's very likely they will string words together that are similar to what's in that material. They sometimes tell lies, though of course they don't "know" that either, because the material they were trained on contains enough false statements to increase the probability of getting off on a wrong track in even one of their next-word predictions.
Obviously, there's more to it all than that, but I believe the complete inability to judge whether a statement is true or false underlies these so-called "hallucinations". They are not closet Nazis trying to make up nonsense to throw us all off. This also explains why they don't say, "Hmmm. Good question. I don't know the answer." They don't know the answer, but they also don't know that they don't know the answer.
Senator Adam Schiff's speech to the Senate was excellent, a speech for the ages; clear, precise, and in the end, there is hope! I will be using parts of the transcript for messaging.
A friend was just telling me how much he enjoys using ChatGPT, and he's the second friend to say so. I tried it for the first time just the other day, but shortly went back to regular Bing search. You have dissuaded me from ever bothering again, for so many reasons, right up there with hallucinations, the ENERGY it takes to power this sly "service."
I agree. I do not trust AI at all because in Google searches when I check the underlying sources, they often do not match the conclusions. AI seems to give you the words it thinks you want, rather than facts.
This is a cool idea. Passing it along so that Robert Hubbell can help promote it if he sees fit. Can someone email it to him?
In less than 24 hours since announcing our plan to launch Liberty Vans equipped with chaplains, lawyers, veterans, and camera crews to follow ICE raids in real time and expose them to the world, we’ve seen an outpouring of support from every corner of the country.
We’ve had supporters step up from ALL 50 STATES. People are ready to fight back and shine a light on the Trump administration's abuse of our country.
We set a goal to raise $500,000 to fund the vans, the crews, the tech, and the training to hold these agents accountable. And thanks to supporters like you, we’re on our way. 𝗕𝘂𝘁 𝘄𝗲’𝗿𝗲 𝗻𝗼𝘁 𝘁𝗵𝗲𝗿𝗲 𝘆𝗲𝘁.
I have an idea. How about instead of continuing to lament the uselessness of legacy and MS media (I know, easy for me to say since I’ve never subscribed to NYT nor WAPO), we could collectively highlight the emerging media sources that are getting the headlines and the coverage right?
It seems far more fruitful to help build what we do want, and walk away from what is no longer serving democracy. Just sayin’.
Robert, a very important part of why I read you almost daily is your humanity. Your intelligence and diligence are important reasons to be sure, but sharing your humanity with your audience is such a relief in this environment of unrelenting bullshit and fundraising. Thank you, and stay away from those chatbots!
In view of the chaos since inauguration, I think the Orange Menace is a sleeper spy for Russia. Maybe Russel Vought too.
As always, THANKS!!
I can't thank you enough for an overview of AI and large language models. I won't use them and now I know why I will continue to resist using them. I'm glad that I retired as university professor in 2018 before all this took off. I don't know how teachers and professors are coping with this access by students to AI.
I just need to know what happened to the esteemed NYT- they excoriated Biden within an inch of his life and they give this joker a pass on everything WTF
As a computer scientist but not an AI researcher, I humbly suggest the reason AI bots "hallucinate" -- an unfortunate word; "lie" is more appropriate -- is that they are programmed to be good at generating natural-language sentences. These are strings of words that are grammatically correct. However, they do not possess a "model of the world" against which to judge the truth of those sentences. Ask your favorite bot, "Do you have a model of the world against which to judge the truth of your statements?" Humans have such models of the world. (It would be nice if some people didn't have such warped models of the world, but such is life, and such is the state of education. :-))
So, bots sometimes tell the truth, though of course they don't "know" that. The material they were trained on contains predominantly true statements, which means it's very likely they will string words together that are similar to what's in that material. They sometimes tell lies, though of course they don't "know" that either, because the material they were trained on contains enough false statements to increase the probability of getting off on a wrong track in even one of their next-word predictions.
Obviously, there's more to it all than that, but I believe the complete inability to judge whether a statement is true or false underlies these so-called "hallucinations". They are not closet Nazis trying to make up nonsense to throw us all off. This also explains why they don't say, "Hmmm. Good question. I don't know the answer." They don't know the answer, but they also don't know that they don't know the answer.
MAGA is grounded in an upside-down world in which good things are bad: compassion, truth, respect for law, the Constitution and other people.
Bad things are good: cruelty, lies, selfishness, bigotry, the Confederacy (Slavery is no stain on our history)
Trump should be king, so when laws and the Constitution get in his way, ignore them.
Senator Adam Schiff's speech to the Senate was excellent, a speech for the ages; clear, precise, and in the end, there is hope! I will be using parts of the transcript for messaging.
A friend was just telling me how much he enjoys using ChatGPT, and he's the second friend to say so. I tried it for the first time just the other day, but shortly went back to regular Bing search. You have dissuaded me from ever bothering again, for so many reasons, right up there with hallucinations, the ENERGY it takes to power this sly "service."
I agree. I do not trust AI at all because in Google searches when I check the underlying sources, they often do not match the conclusions. AI seems to give you the words it thinks you want, rather than facts.
Thank you!
This is a cool idea. Passing it along so that Robert Hubbell can help promote it if he sees fit. Can someone email it to him?
In less than 24 hours since announcing our plan to launch Liberty Vans equipped with chaplains, lawyers, veterans, and camera crews to follow ICE raids in real time and expose them to the world, we’ve seen an outpouring of support from every corner of the country.
We’ve had supporters step up from ALL 50 STATES. People are ready to fight back and shine a light on the Trump administration's abuse of our country.
We set a goal to raise $500,000 to fund the vans, the crews, the tech, and the training to hold these agents accountable. And thanks to supporters like you, we’re on our way. 𝗕𝘂𝘁 𝘄𝗲’𝗿𝗲 𝗻𝗼𝘁 𝘁𝗵𝗲𝗿𝗲 𝘆𝗲𝘁.
𝗪𝗶𝗹𝗹 𝘆𝗼𝘂 𝗰𝗵𝗶𝗽 𝗶𝗻 𝘁𝗼𝗱𝗮𝘆 𝘁𝗼 𝗵𝗲𝗹𝗽 𝘂𝘀 𝗯𝘂𝗶𝗹𝗱 𝘁𝗵𝗲 𝗳𝗶𝗿𝘀𝘁 𝘄𝗮𝘃𝗲 𝗼𝗳 𝗟𝗶𝗯𝗲𝗿𝘁𝘆 𝗩𝗮𝗻𝘀? save-am.org/liberty-vans-v2?t=PjQs7M
-Steve Schmidt, Save America Movement
Text STOP to quit
I have an idea. How about instead of continuing to lament the uselessness of legacy and MS media (I know, easy for me to say since I’ve never subscribed to NYT nor WAPO), we could collectively highlight the emerging media sources that are getting the headlines and the coverage right?
It seems far more fruitful to help build what we do want, and walk away from what is no longer serving democracy. Just sayin’.
Dis and mis information of all kinds. It is maddening. Could you please discuss how much influence Russia has on Fox news?
These are not times for the faint of heart.
The irony is that humans keep inventing things that are dangerous and hard to control.
Why are so many people denying this data?
https://youtu.be/C4x-F5XdCBY?si=ud3HXQvpAJp5nkm-
Robert, a very important part of why I read you almost daily is your humanity. Your intelligence and diligence are important reasons to be sure, but sharing your humanity with your audience is such a relief in this environment of unrelenting bullshit and fundraising. Thank you, and stay away from those chatbots!