#115 - Serial failures
Greg Taieb · · 7 min read
Hello,
Another busy week with news placed on the theme of failure:
- The semi-failure of Apple's bank card
- The complete failure of a lawyer who used ChatGPT
- The fiasco of the management of riots by social networks
- The catastrophic week of Twitter
And otherwise we also talk in video about the impact of Generative AI and audio of the best gadgets of the summer.
💳 Goldman Sachs would already be fed up with the Apple Card, American Express ready to take over
Goldman Sachs, the banking juggernaut which partnered with Apple for the Apple Card, plans to slam the door! Yes, the Wall Street Journal revealed that the big bank would have suffered a loss of a billion dollars, which made them think twice about their alliance with the bitten apple. Don't panic, however, American Express is reportedly in discussions with Apple to take over.
The Apple Card, launched in 2019, was the perfect merger between Apple and Goldman Sachs to create a top-notch credit card experience. It was super easy to use, super secure and even offered you cash rewards. But hey, obviously, it wasn't as juicy for Goldman Sachs as they hoped.
So, we're impatiently waiting for what's next! What is certain is that the explosive mix between tech and finance still has some surprises in store for us.
👩🏻💼Why lawyers should be wary of ChatGPT ?
Some lawyers in the United States have been caught red-handed using completely bogus case law citations. And guess who is behind all this? Our dear ChatGPT, the generative AI that has been talked about so much lately.
Apparently, these unscrupulous lawyers thought they could fool everyone by including fabricated legal references. A team of astute investigators uncovered the deception and these lawyers were ordered to pay a $5,000 fine. It's a clear reminder that technology can be misused, even in the legal field.
This story raises interesting questions about accountability for the use of generative AIs in the real world. Where is the line between AI assistance and deliberate manipulation of information? How can we guarantee integrity and ethics in areas as sensitive as the legal system?
Some specialized services like LegiGPT are trying to develop generative AI solutions based on a legal corpus.
However, having tested it with a lawyer friend, it is clear that the result can still be improved (when it is not simply false).
In any case, this incident underlines the importance of increased vigilance and verification of sources, even when we we face apparently credible references. And for ChatGPT, it is clear that his writing talents can be used for better or for worse.
💥 Digital communication: AI takes power?
Last week I participated alongside Olivier LASCAR (Editor-in-chief of the digital division - Sciences et Avenir) and Vincent MATHIS (Digital consultant & UX/UI designer at Homopixelus) at an edition of the Bal Digital de CanalChat, dedicated to the impact of generative AI in communication.
Exchange as often very enriching thanks to the experts selected by CanalChat (I am of course talking about Olivier and Vincent, I was the tourist)
If you want to watch the replay, it's here 👇
🔥 The role of social networks in the riots
It won't have escaped your notice, the last few days have been very hot in France and it's impossible for me - who shares the news with you on social media - to ignore the subject this week.
To start on a "positive note", the scene which led to the death of young Nahel was filmed and even if the audio still remains subject to interpretation, the broadcast of the images made it possible to have a different point of view from that initially put forward by the police and visibly not 100% clean (I speak with caution because the investigation is still ongoing).
On the other hand, what is beyond doubt is the role of a body of reasoning and amplifiers that social networks played in the nights of the riots.
Between the "humorous" videos Lidl pillagers and the Telegram loops which organize the insurrections, there was an open bar side which really got me posed a problem.
By Camille Lepeintre
Clearly the platforms did not play the game by letting too much content pass that posed problems even though they violated the conditions of use. Tiktoks and Snap and Meta with their video content and their virality are a perfect opportunity for thugs to put themselves on the scene.
So much so that the government had to give them a little boost pressure, to make the platforms face their responsibility. Rumors have even gone so far as to mention plans to cut off networks or even the Internet in certain neighborhoods, which I believe would have represented a first - and a worrying precedent.
But as if all this were not enough, we have also witnessed...
A crazy week at Twitter!
It all started on July 1, when many Twitter users revealed that they had received an error message indicating that they had exceeded the limit on the number of tweets.
At first everyone thought it was a bug except that it was not, as Elon himself told us.
To address extreme levels of data scraping & system manipulation, we’ve applied the following temporary limits:
— Elon Musk (@elonmusk) July 1, 2023
- Verified accounts are limited to reading 6000 posts/day
- Unverified accounts to 600 posts/day
- New unverified accounts to 300/day
Basically as was the case at Reddit a few days ago, Twitter limits access in particular against companies that train their AI engines.
Problem: Twitter is not Reddit and its model is based precisely on advertising and the time we spend there. However, once the limit is reached, we quickly withdraw from the network because we can no longer do anything about it.
Worse, the fact that the tweets can no longer be integrated and readable (I don't understand how the one located just above is displayed) can pose a huge visibility problem for Twitter by limiting the indexing of content from Google.
In any case I know one who did not miss the opportunity to formalize the launch date of his alternative to Twitter: Mark Zuckerberg.
Between 2 training sessions for the next fight in the octagon with Elon, Meta announced the launch date of Threads as July 6 (apparently first in the US).
We even get a short video of the app in this Reuters article.
🪁 What are the Tech gadgets of summer 2023?
With Remi we held our now traditional episode of Encore Un Podcast Tech dedicated to our selection of gadgets and applications for this summer.
We're going to talk about e-readers, drones, connected glasses and of course the best apps for traveling.
I'm quite proud of myself since at the end of the episode, Remi went ahead and ordered one of my dear recommendations (or dear recommendations 🤑).
And that's it for this week!
If this newsletter interested you you can: