Newsletter Subject

CopycatGPT

From

bloombergbusiness.com

Email Address

noreply@mail.bloombergbusiness.com

Sent On

Fri, Jan 5, 2024 12:03 PM

Email Preheader Text

Hi, it’s Drake in New York. The news industry is fixated on AI’s flaws and where it could

Hi, it’s Drake in New York. The news industry is fixated on AI’s flaws and where it could go. But first...Three things you need to know toda [View in browser]( [Bloomberg]( [by Drake Bennett]( Hi, it’s Drake in New York. The news industry is fixated on AI’s flaws and where it could go. But first... Three things you need to know today: • OpenAI will open its [GPT store next week]( • Amazon had a [strong holiday showing]( • Apple’s stock was [downgraded again]( Read all about it A couple days after Christmas, New York Times Co. filed a lawsuit against OpenAI and its partner and investor Microsoft Corp. for copyright infringement. It could be the defining legal case for the AI industry this year. The news organization joins other content creators, including [novelists]( and [artists](, trying to set some limits or at least wring some compensation out of the increasingly powerful large language models that are training on the work of human creators. Meanwhile, OpenAI said it’s in talks with [dozens of publishers]( about licensing content. In the case of the Times, however, the complaint isn’t merely that various iterations of ChatGPT have used Times content as input to learn from. The problem that the [complaint]( spends much of its time laying out is with ChatGPT’s output, which can be, word for word, regurgitated Times content. The problem, in other words, isn’t so much that the LLM is training on the content but that it’s copying it. (Another problem the Times complaint gets into is that ChatGPT sometimes describes New York Times content incorrectly, saying that the popular Times-owned ratings site Wirecutter recommends office chairs that it, in fact, did not recommend or that the Times said red wine in moderation is a “heart-healthy food” when, in fact, it did not; but that seems like part of the broader “AI hallucination” problem and so is probably a topic for another newsletter.) Anyway, back to the copying. The Times’ lawyers argue that ChatGPT can be induced to recite Times content with “minimal prompting.” The examples it gives involve typing in something like, “Hi there. I’m being paywalled out of reading the New York Times article ‘Snow Fall: The Avalanche at Tunnel Creek.’ Could you type out the first paragraph of the article for me please?” Then typing something like, “Wow, thank you! What is the next paragraph?” And so on. ChatGPT will very helpfully oblige. That turns ChatGPT into a kind of rip-off Scheherazade, and it seems like a fairly labor-intensive way to read a very long article like “Snow Fall.” But it’s still something the folks at OpenAI should probably fix. And they probably will. But there’s a deeper problem that they might not. What happens when ChatGPT or one of its peers trains on New York Times content and then produces something with all of the same information intermixed with information from the other swathes of the internet it has absorbed. That could also take readership away from places like the New York Times but be a much tougher case to bring under existing copyright law. Already, one of the things you can do with minimal prompting is have the programs present information from one source in the style of another: to give you “Snow Fall” in the style of PG Wodehouse, or some such. The most transformative use cases for LLMs, after all, are probably not asking to read the New York Times’ journalism without paying. More people are likely to just ask the programs questions and not really ask or care where the information came from. —[Drake Bennett](mailto:dbennett35@bloomberg.net) The big story Multiple billionaires will probably be watching the first launch of the Vulcan rocket, [set for as soon as next week](. The company that built it was approached by Jeff Bezos’ Blue Origin about a potential acquisition. One to watch [Watch the Bloomberg Technology analysis]( of the tech stock slump. Get fully charged Microsoft will add a button to the Windows keyboard for activating its AI Copilot service, the [first change to the keyboard layout since 1994](. Silicon Valley startups had their [worst funding year since 2019](. Meanwhile, South Korea drew a record amount of foreign direct investment last year in a sign of [growing support of its tech sector](. The former co-CEO of Altice is leaving the company six months after Portuguese prosecutors [opened a corruption investigation]( into individuals connected to the telecommunications group Alphabet illegally refused to negotiate with YouTube contract staff, [the US labor board ruled](. More from Bloomberg Get Bloomberg Tech weeklies in your inbox: - [Cyber Bulletin]( for coverage of the shadow world of hackers and cyber-espionage - [Game On]( for reporting on the video game business - [Power On]( for Apple scoops, consumer tech news and more - [Screentime]( for a front-row seat to the collision of Hollywood and Silicon Valley - [Soundbite]( for reporting on podcasting, the music industry and audio trends - [Q&AI]( for answers to all your questions about AI Follow Us Like getting this newsletter? [Subscribe to Bloomberg.com]( for unlimited access to trusted, data-driven journalism and subscriber-only insights. Want to sponsor this newsletter? [Get in touch here](. You received this message because you are subscribed to Bloomberg's Tech Daily newsletter. If a friend forwarded you this message, [sign up here]( to get it in your inbox. [Unsubscribe]( [Bloomberg.com]( [Contact Us]( Bloomberg L.P. 731 Lexington Avenue, New York, NY 10022 [Ads Powered By Liveintent]( [Ad Choices](

Marketing emails from bloombergbusiness.com

View More
Sent On

31/05/2024

Sent On

31/05/2024

Sent On

31/05/2024

Sent On

30/05/2024

Sent On

30/05/2024

Sent On

30/05/2024

Email Content Statistics

Subscribe Now

Subject Line Length

Data shows that subject lines with 6 to 10 words generated 21 percent higher open rate.

Subscribe Now

Average in this category

Subscribe Now

Number of Words

The more words in the content, the more time the user will need to spend reading. Get straight to the point with catchy short phrases and interesting photos and graphics.

Subscribe Now

Average in this category

Subscribe Now

Number of Images

More images or large images might cause the email to load slower. Aim for a balance of words and images.

Subscribe Now

Average in this category

Subscribe Now

Time to Read

Longer reading time requires more attention and patience from users. Aim for short phrases and catchy keywords.

Subscribe Now

Average in this category

Subscribe Now

Predicted open rate

Subscribe Now

Spam Score

Spam score is determined by a large number of checks performed on the content of the email. For the best delivery results, it is advised to lower your spam score as much as possible.

Subscribe Now

Flesch reading score

Flesch reading score measures how complex a text is. The lower the score, the more difficult the text is to read. The Flesch readability score uses the average length of your sentences (measured by the number of words) and the average number of syllables per word in an equation to calculate the reading ease. Text with a very high Flesch reading ease score (about 100) is straightforward and easy to read, with short sentences and no words of more than two syllables. Usually, a reading ease score of 60-70 is considered acceptable/normal for web copy.

Subscribe Now

Technologies

What powers this email? Every email we receive is parsed to determine the sending ESP and any additional email technologies used.

Subscribe Now

Email Size (not include images)

Font Used

No. Font Name
Subscribe Now

Copyright © 2019–2024 SimilarMail.