Hello, I appreciate your in-depth take on Twitter and the information crisis. I'd like to note one thing: your reference for the statement that Twitter is a main source of news and information includes data solely from the US. It's a reference to a Twitter post, and the company is to blame for such narrow focus. There's a big world beyond US borders, where the structure of information sources might - or might not - look differently.
I know this might sound like nitpicking, but we need to do our best to complicate the picture at planetary scale. This is the scale at which Twitter functions, and it includes multiple public spaces, some very different from the US public space. This is especially relevant that norms (free speech!), metaphors, implicit biases that are common in this space are constantly pushed onto the rest of the world.
Google+ used to have a feature called Collections. The idea behind it was that users would categorize their content into different buckets, i.e. politics, sports, memes, or whatever. That allowed their followers to subscribe or unsubscribe from individual collections based off of their interest rather than having to completely unfollow someone if you hated reading their posts about sports or something. I’ve often thought that the entire thing was a strategy by Google to get thousands of people to train their algorithms by having them manually categorize text and photos. Regardless, it did provide for a great end user experience.
Agree, that's probably more effective for a better user experience. To keep users engaged more time, maybe not, I'm not sure. But if anything, it's not a tech limitation the reason why companies aren't implementing these types of features.
Yeah, I think it’s all about ads. I don’t know for sure, but that’s my guess. My thinking is that not showing you all of the content that you want and know exists causes you to spend more time on the platform which allows them to raise ad rates because of increased engagement times.
I’ve always thought that there was an opportunity for social networks to use large language models and ML as a way to identify specific topics that people were talking about. Then, if the platform knows that you’re interested in cars or retro computers or whatever it could easily show you that content. And yet, no one seems to be leveraging this. It seems obvious to me. Is there a reason no one has done this? Is it because of the adversarial nature that ads introduce to these platforms or is there some technical issue that limits implementation?
Hi John! That's exactly how TikTok's recommender algorithm works (other social media apps use similar systems, too). However, there's no need for language models in this case--TikTok simply shows you more of what you watch longer and repeatedly.
In which ways do you think language models can provide an advantage here? I'm curious!
Because I primarily hang out on text-heavy platforms. Twitter could tell I was reading someone’s tweets, but most of the time I’m reading them because of the topic as opposed to because I’m interested in that specific person. That’s why I think LLMs could provide an opportunity for platforms because it allows them to step away from the idea of “John read person X’s tweets” so we’ll show him more of that. It’s quite possible that I love their tweets about computer science, but find their politics to be odious. It seems like LLMs could be leveraged to figure out that I’m interacting with and responding to tweets about a specific topic as opposed to just tweets from person X.
I suppose TikTok does this in some way because they are able to figure out what types of videos you’re watching. I’ve never used TikTok so I don’t know if that’s because of user-entered hashtags or if it’s just machine learning doing it all.
Oh, that makes a lot of sense. TikTok uses ML, yes, and it doesn't need direct user reports (likes, comments, hashtags, etc.), it's more implicit, it uses user's implicit behavior like watch times.
Could Twitter do this? Maybe. Do we want it? That's another question...
I get that Twitter is broken, but wouldn’t we want every platform to do this? If I’m enjoying something it seems to make sense that I’d want more of it. That includes political fighting. There are some people who genuinely enjoy that. There are two sides to this. When the platform is showing me things I want it’s also not showing me things I don’t want. Of course, if someone is prone to believing conspiracy theories or has begun a journey that leads to radicalization then it’s going to cause societal problems. That’s a problem that still needs to be solved or addressed in some way.
Here's the thing: content you want to see =/= content you'll engage with.
Social media platforms aren't incentivized to just show you what you want to see, but what'd keep you online--even if you dislike it and stay because of hatred, fear, anxiety, etc.
And it's not that Facebook or TikTok do this on purpose. They want you to stay at all costs and the algorithm itself finds what makes you stay. The companies simply don't care what kind of content that would be.
Hey Alberto, would you consider writing an article about this topic? The idea that many times AI will fulfill our requests/complete our stated goals on ways that we don’t expect and sometimes don’t want?
This man is an opportunist. In that sense, he is a branding specialist with charisma, decent looks, and an authoritarian background. He understands engineering, but he did not invent electric cars or batteries or the technology to run such. He did not invent space travel. Governments did. He gets people to the space station and back to do a job alive. That is good, and it is the right time for this. Due to climate issues, it is the right time for electric cars. He makes such cars for rich people. Is it the right time to buy twitter, become authoritarian with it, and charge cold hard cash to play? Maybe. As for AI, it is a great tool in the right hands at the right time. Scientists and medical professionals are making excellent use of AI and will continue to do so. Will Musk use it well? We will see. He is a good brander. He does have charisma. People do his bidding. But, AI is in the past, always subject to data as it is processed in. Twitter operates in nanosecond present time. Can AI?
I also wonder if this is dumming people down. We give them what they want but nothing else. We do this to sell to them. We don't challenge them to think outside the box the algorithm is defining.
I don't know in which ways it could make us dumber, but I certainly feel that social media giving us exactly what is needed to keep us online is causing important problems.
Hello, I appreciate your in-depth take on Twitter and the information crisis. I'd like to note one thing: your reference for the statement that Twitter is a main source of news and information includes data solely from the US. It's a reference to a Twitter post, and the company is to blame for such narrow focus. There's a big world beyond US borders, where the structure of information sources might - or might not - look differently.
I know this might sound like nitpicking, but we need to do our best to complicate the picture at planetary scale. This is the scale at which Twitter functions, and it includes multiple public spaces, some very different from the US public space. This is especially relevant that norms (free speech!), metaphors, implicit biases that are common in this space are constantly pushed onto the rest of the world.
You're totally right. I'll edit it to make it clear.
Google+ used to have a feature called Collections. The idea behind it was that users would categorize their content into different buckets, i.e. politics, sports, memes, or whatever. That allowed their followers to subscribe or unsubscribe from individual collections based off of their interest rather than having to completely unfollow someone if you hated reading their posts about sports or something. I’ve often thought that the entire thing was a strategy by Google to get thousands of people to train their algorithms by having them manually categorize text and photos. Regardless, it did provide for a great end user experience.
Agree, that's probably more effective for a better user experience. To keep users engaged more time, maybe not, I'm not sure. But if anything, it's not a tech limitation the reason why companies aren't implementing these types of features.
Yeah, I think it’s all about ads. I don’t know for sure, but that’s my guess. My thinking is that not showing you all of the content that you want and know exists causes you to spend more time on the platform which allows them to raise ad rates because of increased engagement times.
Maybe there's another take on this. Twitter is already irrelevant so these AI-powered bots spewing misinformation is just noise. Ad revenue will find another suitable, stable place soon. Charlie Warzel used the term "Geriatric Social Media" (https://newsletters.theatlantic.com/galaxy-brain/63609043b606fe00376a82da/welcome-to-geriatric-social-media/).
Thanks for the link Sharif, interesting read!
I’ve always thought that there was an opportunity for social networks to use large language models and ML as a way to identify specific topics that people were talking about. Then, if the platform knows that you’re interested in cars or retro computers or whatever it could easily show you that content. And yet, no one seems to be leveraging this. It seems obvious to me. Is there a reason no one has done this? Is it because of the adversarial nature that ads introduce to these platforms or is there some technical issue that limits implementation?
Hi John! That's exactly how TikTok's recommender algorithm works (other social media apps use similar systems, too). However, there's no need for language models in this case--TikTok simply shows you more of what you watch longer and repeatedly.
In which ways do you think language models can provide an advantage here? I'm curious!
Because I primarily hang out on text-heavy platforms. Twitter could tell I was reading someone’s tweets, but most of the time I’m reading them because of the topic as opposed to because I’m interested in that specific person. That’s why I think LLMs could provide an opportunity for platforms because it allows them to step away from the idea of “John read person X’s tweets” so we’ll show him more of that. It’s quite possible that I love their tweets about computer science, but find their politics to be odious. It seems like LLMs could be leveraged to figure out that I’m interacting with and responding to tweets about a specific topic as opposed to just tweets from person X.
I suppose TikTok does this in some way because they are able to figure out what types of videos you’re watching. I’ve never used TikTok so I don’t know if that’s because of user-entered hashtags or if it’s just machine learning doing it all.
Oh, that makes a lot of sense. TikTok uses ML, yes, and it doesn't need direct user reports (likes, comments, hashtags, etc.), it's more implicit, it uses user's implicit behavior like watch times.
Could Twitter do this? Maybe. Do we want it? That's another question...
I get that Twitter is broken, but wouldn’t we want every platform to do this? If I’m enjoying something it seems to make sense that I’d want more of it. That includes political fighting. There are some people who genuinely enjoy that. There are two sides to this. When the platform is showing me things I want it’s also not showing me things I don’t want. Of course, if someone is prone to believing conspiracy theories or has begun a journey that leads to radicalization then it’s going to cause societal problems. That’s a problem that still needs to be solved or addressed in some way.
Here's the thing: content you want to see =/= content you'll engage with.
Social media platforms aren't incentivized to just show you what you want to see, but what'd keep you online--even if you dislike it and stay because of hatred, fear, anxiety, etc.
And it's not that Facebook or TikTok do this on purpose. They want you to stay at all costs and the algorithm itself finds what makes you stay. The companies simply don't care what kind of content that would be.
Hey Alberto, would you consider writing an article about this topic? The idea that many times AI will fulfill our requests/complete our stated goals on ways that we don’t expect and sometimes don’t want?
Very good point.
Because, these ads are becoming useless. Like a subprime bubble. Tim Hwang has a great book on this topic. "Subprime Attention Crisis
Advertising and the Time Bomb at the Heart of the Internet" (link: https://us.macmillan.com/books/9780374538651/subprimeattentioncrisis).
Thanks for the book recommendation. I grabbed it and look forward to reading it.
This man is an opportunist. In that sense, he is a branding specialist with charisma, decent looks, and an authoritarian background. He understands engineering, but he did not invent electric cars or batteries or the technology to run such. He did not invent space travel. Governments did. He gets people to the space station and back to do a job alive. That is good, and it is the right time for this. Due to climate issues, it is the right time for electric cars. He makes such cars for rich people. Is it the right time to buy twitter, become authoritarian with it, and charge cold hard cash to play? Maybe. As for AI, it is a great tool in the right hands at the right time. Scientists and medical professionals are making excellent use of AI and will continue to do so. Will Musk use it well? We will see. He is a good brander. He does have charisma. People do his bidding. But, AI is in the past, always subject to data as it is processed in. Twitter operates in nanosecond present time. Can AI?
Well, Twitter already uses AI for the recommender algorithm that decides what ends up in our feeds.
About Musk, true, he has charisma and is an opportunist. But decent looks... ;)
Have twitter's recommender systems worked well in the past? I can say that Facebook uses them and what they recommend to me is boring or irritating.
Well, I don't think recommender systems work perfectly, but they do work, that's for sure (otherwise social media wouldn't be so addictive!)
I also wonder if this is dumming people down. We give them what they want but nothing else. We do this to sell to them. We don't challenge them to think outside the box the algorithm is defining.
I don't know in which ways it could make us dumber, but I certainly feel that social media giving us exactly what is needed to keep us online is causing important problems.
Maybe. Many folks are amused easily. I work with a man who teaches AI and he thinks them low brow and ho hum.