I like free!
20 characters
1 Like
I thought using BingChat in creative mode uses GPT4.
Do you like having your data packaged up and sold to advertisers too? 
No, which is why I use DDG as my default, ad blockers, and even VPNs. 
But I wonder if we can escape it using other forms of AI? I have no idea. If/when AI becomes more reliable, it may be one of the few services I’m willing to pay a decent price for, and I hope that by paying, I’m not selling myself and my data.
Just curious: Charles Francis Adams would not be noted in the citation? Also, do you go to the original source to double check the quote?
AFAIK it uses GPT4 for all modes.
If you move a Mouse, or your Fingers moving across a TrackPad, you get a reliable Response.
If you use the technic called today “AI” you are using a regular computer program, with all its limitations, but with the expectation that you are using something from an other world, or an other time! This is not a reliable Response.
If you don’t want to cheat or produce some kind of fakes, there is no real value you could get out of those “AI” today (maybe beside some code snippets for developers and BlackHats, and an increase in readability for Phishing-Mails from some criminals), as there is still no “Intelligence” among them, as the users are expecting, and getting a valuable result remain pure luck!
Those systems are also not “learning” in a way the normal user is expecting from them, as you could pretty nicely see with the different responses ChatGPT produced today for the same question.
Those labels “AI” are just some Marketing Fakes, and I really, really do not like Fakes!
We already are having enough problems in the “Western World”, we do not need those additional sources for trouble in our live!
I added that information in EndNote to make it more accurate, ChatGPT did not include it. I’m still double-checking for original sources. My purpose in using AI was to jumpstart and speed up the initial search so I can ultimately find the original source.
1 Like
I am software engineer by trade and damn I find Shortcuts completely enigmatic. A don’t get me started with AppleScript!
1 Like
I do not think @Bmosbacker was under that expectation. Do you believe all the media hype you read? Neither do the rest of us.
Well that’s for the users to decide. And obviously in this case there was real value.
Perhaps you are approaching it from the wrong use case. It’s not Google, it’s not an all-knowing entity. I have used it --not much as I would like-- in my professional work some times and it just pays for the subscription itself. And yes, I’ve got my fair share of stupid and wrong answers --try making it create regexps!
What troubles are causing LLMs to you, exactly? While I agree that yes, we have enough problems, it’s not like LLMs will end the civilization.
3 Likes
Suddenly I feel a whole lot smarter, or at least less ignorant. 
1 Like
I guarantee I’m not. “This is a really cool algorithm which, based on its training data, can put together coherent sentences and even complete arguments.” That’s what it is. And the fact that it doesn’t necessarily return the same result from the same input makes it more valuable, not less.
Go track down quite literally any human, and ask them to explain something to you. They might give you the same response each time, but the odds are good you’ll get at least somewhat different results on different days. The fact that the AI can return more than one result for a given query - as long as the result is actually relevant to the query - makes it more like a human, not less. IMHO, anyway.
In the examples above, the AI didn’t fail - it just gave a different, arguably-correct answer to the question.
Oddly enough, Google’s algorithm doesn’t return the same results for the same query either. Take a user in Germany and a user in the USA, give Google the same query, and it will yield different results because it knows which results Germans prefer vs. Americans. And based on your entire previous search history, it has some idea what you probably want - so it can skew the results accordingly.
If you follow Seth Godin, a ways back he took his old audio archives and fed them to some sort of modeling algorithm, then had the algorithm voice his podcast. And he fed his blog archives to a chatbot, which you can ask questions of - Ask an AI trained on Seth’s writing | Seth's Blog
It takes awhile to answer, but it links up old posts and prepares a response based on your question.
Oddly enough, this sort of thing is literally the ultimate goal of the whole “building a second brain” movement. You could train an LLM based on all of your old writing, then have a conversation with yourself. 
1 Like
This is far more appealing to me than the existing publicly available LLMs. A large language model that helps me navigate all the stuff I have put into it? Yes.
One built from a secretive collection of who knows what from who knows where, with no way of tracing where it actually came from?* No thanks.
*Because remember, it’s compiled from many sources; I don’t think the “source” they give when you ask for attribution is anything of the sort; it’s just the most likely collection of words associated with the material and terms relating to sourcing.
2 Likes
This thread is interesting because over the last couple of years I’ve tried without success to trace a quote I read once that I foolishly didn’t write down (never read a quote, think “ha that’s nice”, and then NOT write it down!!). Google hasn’t helped, and actually I spent sometime on my quest again this morning after seeing this thread, once again without success (and let’s be honest, if the correct answer hasn’t got to pages 1-3 of Google, Google hadn’t succeeded). I’m now wondering if ChatGPT would have more success.
Twitter used to be great for this stuff back in its heyday. Reddit probably still is if you ask in the right place. You could throw out a random question about a book with a yellow cover that had a funny comment about a cat, and someone would usually be back to you with the correct answer. I’ve retraced several books that way, and songs, which can be even harder to trace given how fragmented our memories can be. The Twitter hivemind once traced a song with only the comment that it had a sad piano intro, was sung by a man with a deep voice and I thought the first line started with a man’s name. Even I thought I was being unreasonable, but I’d spent several weeks trying to find the annoying earworm, and two people came back to me with the right answer within an hour. To me it remains Twitter’s greatest feat
LibraryThing recently released a ChatGPT integration on their site that will attempt to trace mystery books for you based on whatever details you can remember. I’ve not tested it.
Valid topics, for sure but as you are mentioning, if I correctly understand you think that generative AIs are qualitatively more dangerous because they enable the proliferation of deep fakes.
I have a certain personal hope which may be absurd given the state the world is in, but there you go: deepfakes will make us become more critical of what is being fed to us in the daily news rotation and we will resort to media sources that bet their reputation and in the end their business model on fact checking.
But the cat is already out of the box --as societies we need to do that even before general AI deployments.
Ulli, I welcome and find value your incandescent criticism. Please consider using an extra line break in between your paragraphs.
Best,
01001001
Given the discussion above I thought I’d share this quote from a recent The Atlantic article:
That future is now here. Did you see the recent photos of NYC police officers aggressively arresting Donald Trump? Or of the pope in a puffer jacket? Thanks to AI, it takes no special skills and no money to conjure up high-resolution, realistic images or videos of anything you can type into a prompt box. As more people familiarize themselves with these technologies, the flow of high-quality deepfakes into social media is likely to get much heavier very soon.
Some people have taken heart from the public’s reaction to the fake Trump photos in particular—a quick dismissal and collective shrug. But that misses Bannon’s point. The greater the volume of deepfakes that are introduced into circulation (including seemingly innocuous ones like the one of the pope), the more the public will hesitate to trust anything. People will be far freer to believe whatever they want to believe [emphasis added]. Trust in institutions and in fellow citizens will continue to fall.
Haidt, J., & Schmidt, E. (2023, 2023-05-05). AI Is About to Make Social Media (Much) More Toxic. The Atlantic. https://www.theatlantic.com/technology/archive/2023/05/generative-ai-social-media-integration-dangers-disinformation-addiction/673940/
1 Like
YES, it is absurd…!!
I can technically pinpoint the exact source of the issue: it’s the programmatic advertisement machinery that allows any media outlet to target specific audiences with surgeon precision. While it was designed for serving ads in the most effective way, it was obviously used for more obscure purposes.
And talking about cats and boxes, and the reason for hope is the governments response to that with GDPR in the Euro zone which is, among other things, directed against those use cases. Also from the corporate side, Apple is presenting first party privacy as a value proposition, and its consumers are not precisely unhappy about that.
Can generative AIs “generate” the same disgrace? Potentially yes. Have governments been too slow to come up with suitable regulations in the past? Hell yes, technology moves way faster than politicians can legislate. But, again on the bright side, see the Italian privacy watchdog reaction to OpenAI’s developments: “you need to stop while we figure your thing out”. And that’s ok.
No offense taken at all! Again, thanks for your observations.
2 Likes
Okay, I’ve got to check out the pay-per-use API bit. $20/month is incredibly steep, if I can cut that down by building an application, that’s something of value that I will pursue. And I’ll do it with the help of ChatGPT.
I’m an amateur coder. This past weekend, I used ChatGPT to write a NodeJS script that checks the availability of books/audiobooks across four libraries. This little gem will save the time and hassle that I’ve been spending going to the individual websites to do the same check myself. It took me a few hours to coax the code out of ChatGPT, but I enjoyed almost every minute of it - so I had fun AND got value out of the experience.
3 Likes