The Gist

  • New features for ChatGPT. Security and privacy issues addressed with latest reveal.
  • Is Meta catching up on AI? Mark Zuckerberg says his company is "no longer behind."
  • No GPT for you. OpenAI unveils new branding policy. 

In an apparent demonstration of its commitment to improve the ChatGPT user experience, OpenAI has announced several new data management features to give users more customization options and greater control.

In response to security and privacy concerns, users will now have the ability to turn off their chat history within the ChatGPT settings. Once disabled, any new conversations are only retained for 30 days and will only be reviewed if necessary to monitor for abuse. After this period, the conversations will be permanently deleted.

OpenAI researchers are also developing a new subscription service for ChatGPT — called ChatGPT Business. Company officials say the service, tentatively planned for release in “the coming months,” will provide users with increased data management options and greater control over the end-user data within their organization. ChatGPT Business will follow OpenAI’s API data usage policies that state end users’ data won’t be used to train models by default.

Lastly, ChatGPT will now offer a new “Export” option in settings that allows users to export their data and review the information being stored. Once the export is requested, users will receive an email with a file containing all relevant data, including their conversations.

Sounds good.

In other AI news...

Alphabet and Google CEO 'Excited' about AI

During its Q1 2023 earnings conference call on Tuesday, Alphabet placed a notable emphasis on the progress made in AI technology.

In a blog last week, Sundar Pichai, CEO of Alphabet and Google, revealed the creation of a new unit called Google DeepMind, bringing together two leading research groups in the AI field, the “Brain” team from Google Research and DeepMind. The goal is to focus on AI development and building more capable, safe, responsible systems.

During Tuesday’s call, Pichai said he was “quite excited” about bringing “these two world-class teams” together and believed “their collective accomplishments in AI over the last decade have really set the stage for this moment.”

“And so, both getting access to pooled talent so that they can work together in a coordinated way will help us pool our computational resources too, which is going to be critical, and will help us build,” Pichai said. “The core product is obviously building more capable models safely and responsibly and ... taking into account all of the capabilities our customers need, both on the consumer side and the Cloud side and being able to iterate and getting that virtuous cycle going.”

When asked what AI milestones the public should look out for, Pichai pointed to the recent release of PaLM APIs, and said, "We are incorporating PaLM across our products, but we will continue that progress, and we'll keep you posted as we do.”

Related Article: Salesforce Integrates Einstein With Flow, Musk Wants TruthGPT, More AI News

AI Video of the Week: Pushing AI Boundaries

Runway CEO Cris Valenzuela discusses pushing the boundaries of AI and the rise of investor interest with Bloomberg Technology:

Zuckerberg: Meta 'No Longer Behind in Building AI Infrastructure'

Meta CEO Mark Zuckerberg touted AI and efficiency during a chat with investors yesterday in the company’s first quarter earnings call. With $28.6 billion in revenue and a profit of $5.7 billion for the quarter, Zuckerberg called it “a good quarter and our community continues to grow.”

"Our AI work is driving good results across our apps and business,” Zuckerberg said. “We're also becoming more efficient so we can build better products faster and put ourselves in a stronger position to deliver our long-term vision.”

He went on to assure investors that Meta is “no longer behind” in building its AI infrastructure and hinted that new AI product releases will be announced in the months ahead. During the call, Zuckerberg said the company is “exploring chat experiences in WhatsApp and Messenger” in addition to visual creation tools for Facebook and Instagram posts, and in time, “video and multi-modal experiences.”

“I expect that a lot of interest in AI agents for business messaging and customer support will come once we nail that experience,” Zuckerberg said. “Over time, this will extend to our work on the metaverse, too, where people will much more easily be able to create avatars, objects, worlds, and code to tie all of them together.”

The news wasn't all great for Meta. A hiring freeze is currently in place at Meta, and the company reportedly plans to lay off about 11,000 staffers before the end of the year.

OpenAI Unveils New Branding Guidelines

With Elon Musk’s recent announcement of plans to build a ChatGPT rival called TruthGPT, one begins to wonder — just how many products have commandeered GPT? GPT stands for Generative Pre-trained Transformer — and it seems like just about everybody has tacked it onto the end of their own, new AI product.

Learning Opportunities

But no more. OpenAI is in the middle of a trademark application for GPT, but that could take months, notably because while the company did apply for a fast-tracked process, it also forgot to include the application fee resulting in a denial of that request. Instead of waiting for an approval that could take several months, OpenAI released a new set of branding guidelines that govern how GPT can be used in branding and communications. Those who’ve already glommed the GPT surname, may need to think about coming up with their own catchy moniker.

New Study: Can ChatGPT Predict the Stock Market?

 

With all the talk of generative AI replacing human jobs, stockbrokers should take note of this ...

 

A recent study by researchers at the University of Florida found that OpenAI’s ChatGPT was surprisingly accurate in predicting stock market movements. The study used daily returns from the Center for Research in Security Prices (CRSP) and related news headlines from October 2021 through December 2022 to train ChatGPT to predict stock market trends, and then evaluated its accuracy in predicting future stock prices. The study revealed “a strong correlation” between the ChatGPT evaluation and the subsequent daily returns of the sample stocks.

 

The study also revealed that more “basic models” like GPT-1, GPT-2, and BERT did not forecast returns accurately, “indicating return predictability is an emerging capacity of complex models.” Researchers believe the results could mean that ChatGPT may be a “valuable tool for predicting stock market movements based on sentiment analysis.”

AI Tweet of the Week: Artist Offers AI Royalty Split

A Canadian singer named Grimes (real name, Claire Elise Boucher) has offered to split 50% of royalties from “any successful AI generated song” that uses her voice — and don’t forget the “successful” part.

Grimes tweeted the offer out on Sunday to her more than 1.5 million followers after news that “Heart on My Sleeve,” a song reportedly created using AI with replicated vocals from Drake and The Weeknd, received millions of views on TikTok.

But most interesting to me — is her relationship to Elon Musk (yes her Twitter account is verified.) Grimes and Musk were in a relationship until 2021 and have two children together, a son named X AE A-XII and a daughter who Grimes said is referred to as “Y” or “Why” or “?”

fa-regular fa-lightbulb Have a tip to share with our editorial team? Drop us a line: