Profile picture
Claire Berlinski @ClaireBerlinski
, 18 tweets, 4 min read Read on Twitter
It is dangerous, yes, for once I agree with Trump. But he's missing the point. The notable thing isn't the banned accounts. It's the change in algorithms that's made Twitter so much more pleasant recently. uk.reuters.com/article/uk-usa…
I'm sure everyone enjoys that as much as I do, and that it's very good for Twitter as a business decision. But to be able to change the user experience as much as they have in--what, maybe the past month?--is a truly spooky achievement in AI.
Here's an anodyne description of what they're doing: engadget.com/2018/07/30/twi… And very successfully. For users, this is a much better product. But what's spooky about this is that their algorithms can now distinguish, very accurately, between "polite" and "impolite."
It is, obviously, not a human being at Twitter that's deciding who's a troll and whose responses should be hidden from you because they're rude. It's an algorithm. And this algorithm is very good. engadget.com/2018/05/15/twi…
Note that only a year ago, we were being reassured that this was still way beyond AI: theatlantic.com/technology/arc… Ian Bogost writes in that article that the term AI is "meaningless," and uses *this example* exactly.
"Deflationary examples of AI are everywhere. Google funds a system to identify toxic comments online, a machine learning algorithm called Perspective. But it turns out that simple typos can fool it."
Well, they can't fool it anymore. Twitter's system is working way better than you'd expect, particularly if all they've done, as he writes is add, "additional clauses in database queries." I don't think it's quite as simple as that, actually.
"Twitter [is] using machine learning to pick out ... low quality replies," says the article he scorns as an example of "AI inflation." thenextweb.com/twitter/2017/0… But it seems that is indeed what they're doing. Successfully.
What are the implications of that? No one knows. Is this an inflection point? Well, it's something we were laughing at a year ago. So yes, it seems we're getting closer, faster than expected, to what were once unimaginable moral dilemmas. conversationswithbillkristol.org/video/jim-manz…
And we'll be in competition with countries that also have this ability.
I find it a little strange that so few people have noticed a radical change in Twitter. I suspect that's because people prefer to think that they're causing all these positive interactions.
It reminds me of Freud's observations about why his female patients often fell in love with him. His lesser colleague, Otto Breuer, observed the same thing and concluded it must be because he was so lovable. Freud's genius was in his modesty:
The patients' love, he concluded, was induced by the analytic relationship--which allows the patient to perceive in the analyst a transferred imago of her infantile attachments. Her vision of the analyst was a projection.
It would be easy for me to attribute all the positive replies I'm going to see in response to this comment to my own lovability. But only about a month ago, any allusion to Freud on Twitter would have guaranteed I'd see an overwhelmingly hostile and derisory response.
Now, it seems, people who hate Freud are much less apt to see that comment in the first place. If they do, and react the way people who hate Freud usually do, I'm much less apt to see the response. That's *quite* sophisticated.
And it's not because my opinions have suddenly become more lovable or because Freud has been rehabilitated in public opinion. It's because Twitter has figured out what kind of response I find pleasant or interesting--and made the others disappear from my view. And vice-versa.
I appreciate that. It makes me feel more lovable. But it's not a power I want in the hands of my worst enemies, which is how we have to evaluate any new power. So the problem isn't that they've banned Alex Jones (although they shouldn't have--that's liberal democracy 101, @jack.)
It's that AI is becoming more sophisticated, and faster, than many of us expected. Twitter now knows how to show me a more flattering transferred imago. What else could someone do with that ability?
Missing some Tweet in this thread?
You can try to force a refresh.

Like this thread? Get email updates or save it to PDF!

Subscribe to Claire Berlinski
Profile picture

Get real-time email alerts when new unrolls are available from this author!

This content may be removed anytime!

Twitter may remove this content at anytime, convert it as a PDF, save and print for later use!

Try unrolling a thread yourself!

how to unroll video

1) Follow Thread Reader App on Twitter so you can easily mention us!

2) Go to a Twitter thread (series of Tweets by the same owner) and mention us with a keyword "unroll" @threadreaderapp unroll

You can practice here first or read more on our help page!

Did Thread Reader help you today?

Support us! We are indie developers!


This site is made by just three indie developers on a laptop doing marketing, support and development! Read more about the story.

Become a Premium Member and get exclusive features!

Premium member ($3.00/month or $30.00/year)

Too expensive? Make a small donation by buying us coffee ($5) or help with server cost ($10)

Donate via Paypal Become our Patreon

Thank you for your support!