OT: using our data to train AI

[mjd1001]

For those of us who are Twitter users and care about this AI stuff.....

If you’re an X/Twitter user, take a second to tweak this setting. It turns out many user posts have been automatically opted-in for training Grok, Elon Musk’s ChatGPT competitor.

 

Several users, including open-source intelligence analyst Oliver Alexander, noticed the change and alerted the public. “X has now enabled data sharing by default for every user, which means you consent to them using all your posts, interactions and data on here to train Grok and share this data with xAI,” he said in a post on X.

If you don’t want X hoovering up every little interaction you have to train its terminally online large language model, you need to opt-out. Users can only do this from the website, the setting isn’t present on either the Android or iPhone X app.

https://gizmodo.com/how-to-stop-elon-musk-from-training-his-ai-on-your-data-2000479425

Edited July 26 by mjd1001

[matter2003]

Understand people's trepidation about it but it's like trying to take a small garden trowel and dig out from an avalanche.

There are stores now where there are no humans, but the AI sees what you take and then charges you for it based on the information on file with your credit card, etc.  A guy who is a leader in AI was talking about how he was in one of these stores in NYC and he looked at some chocolate bars, decided against it and put them back, then got a drink instead. Walked out and was charged for the drink.

But guess what he started having ads pop up for while he was browsing within the next 15 minutes?  Yup...you guessed it...chocolate.

Edited July 26 by matter2003

[mjd1001]

5 minutes ago, matter2003 said:

Understand people's trepidation about it but it's like trying to take a small garden trowel and dig out from an avalanche.

There are stores now where there are no humans, but the AI sees what you take and then charges you for it based on the information on file with your credit card, etc.  A guy who is a leader in AI was talking about how he was in one of these stores in NYC and he looked at some chocolate bars, decided against it and put them back, then got a drink instead. Walked out and was charged for the drink.

But guess what he started having ads pop up for while he was browsing within the next 15 minutes?  Yup...you guessed it...chocolate.

I'm sure there is a simple explanation, but I have something like that happen to me on a regular basis.

Our daughter calls my wife and talks to her about a new couch she wants to buy for the living room. My wife looks it up on her desktop computer. Doesn't buy anything, just looks at it and loads the picture.  30 minutes later, I LEAVE the house, open up a page on my phone and an ad is there, on MY Phone for the store and the exact same couch my wife looked at earlier.

[matter2003]

4 minutes ago, mjd1001 said:

I'm sure there is a simple explanation, but I have something like that happen to me on a regular basis.

Our daughter calls my wife and talks to her about a new couch she wants to buy for the living room. My wife looks it up on her desktop computer. Doesn't buy anything, just looks at it and loads the picture.  30 minutes later, I LEAVE the house, open up a page on my phone and an ad is there, on MY Phone for the store and the exact same couch my wife looked at earlier.

browsers are probably synced across devices

[LTS]

4 hours ago, mjd1001 said:

I'm sure there is a simple explanation, but I have something like that happen to me on a regular basis.

Our daughter calls my wife and talks to her about a new couch she wants to buy for the living room. My wife looks it up on her desktop computer. Doesn't buy anything, just looks at it and loads the picture.  30 minutes later, I LEAVE the house, open up a page on my phone and an ad is there, on MY Phone for the store and the exact same couch my wife looked at earlier.

There is a very simple explanation actually. Your profiles are linked in a greater identity resolution and marketing database. These companies collect all manner of information from all manner of sources and then link them to your virtual identity. They then use relational data to create circles of connections.  Advertising companies (and many others) then pull this data to help determine ad relevancy and boom.

It can be as simple as your phone unique identifier and your friend's phone unique identifier were in close proximity for a certain amount of time. Your friend did a search for "XYZ" and through relational data the system reasonably assumed you two might have had a conversation about it and thus you see an advertisement on your phone a little bit later.

I think the extent to which our lives can be modeled through collection of data would blow most people's minds.  No doubt people expect the more obvious items but the depth goes far beyond that.

Certainly linking browsers and/or a shared profile on her desktop computer and your browser would do it as well.  The points of connectivity are endless.

[...]

What do you think Google, Microsoft and Facebook are doing? Why do you think ads show up on Google and Facebook that reflect the conversations you're having in your home? That's an AI trained on your specific online behavior and the things you say. Microsoft was open about Windows 11 logging every stroke of the keyboard and swipe of the mouse - who knows what you're agreed to for prior versions of Windows that haven't been disclosed. All of that data is used to train AIs.

Elon at least talks about how Grok is being trained and is against biasing (censoring, filtering) its responses.

[Pimlach]

44 minutes ago, ... said:

What do you think Google, Microsoft and Facebook are doing? Why do you think ads show up on Google and Facebook that reflect the conversations you're having in your home? That's an AI trained on your specific online behavior and the things you say. Microsoft was open about Windows 11 logging every stroke of the keyboard and swipe of the mouse - who knows what you're agreed to for prior versions of Windows that haven't been disclosed. All of that data is used to train AIs.

Elon at least talks about how Grok is being trained and is against biasing (censoring, filtering) its responses.

This is happening way to often. 

I do not have any accounts for FaceBook, X, Instagram, or Tik Tok.  Yet  my PC and my iPad and my iPhone all to seem to know what I am doing, what I am looking at, and even what I was talking about.  

It seems to be an invasion of privacy and incriminating too.   

[PerreaultForever]

AI still has a lot of learning to do. It thinks Peyton Krebs is a dynamic signing for the New York Rangers.

https://www.youtube.com/watch?v=6p8Lzh_La2o&t=125s

Edited July 26 by PerreaultForever

[Big Guava]

1 hour ago, PerreaultForever said:

AI still has a lot of learning to do. It thinks Peyton Krebs is a dynamic signing for the New York Rangers.

https://www.youtube.com/watch?v=6p8Lzh_La2o&t=125s

Has a lot to learn about writing code too.

This nonsense about how AI is going to put all software engineers out of a job within 5 years is complete rubbish.

It's definitely a productivity boost when used in conjunction with a competent software engineer, but if someone is incompetent it's not going to be much use to them or it's going to build you something that either doesn't work or doesn't even do anything close to what it is supposed to do and they won't know better or how to instruct it properly.

I use it everyday at work with GitHub Co-Pilot and it's very good at writing repetitive boilerplate code, documentation and unit tests. Not so great at much else. It will send you down a rabbit hole very quickly if you let it.

Edited July 27 by Big Guava