NOTEWORTHY NEWS: OUR TAKE ON THE LATEST AI/GENAI NEWS

Noteworthy AI News for Agencies - Eighth Edition

By Jeremy Lockhorn , SVP, Creative Technologies & Innovation, 4A's

Please note that Noteworthy News will not be published next week (July 5) as the 4A’s takes time off to celebrate Independence Day. We’ll be back at it the following week with a double-dose of AI news for agencies.

THE BIG STOR(IES)

NEW 4A’S RESOURCES

In case you missed it, we released a couple of members-only reports this week aimed at helping agencies understand the current state of GenAI adoption and plot a course to agency-wide implementation. More details below:

???? The State Of Generative AI Inside US Agencies, 2024?

Based on a recent survey of 4A's members, this piece from Forrester and the 4A’s dives deep into insights on adoption, objectives, use cases, benefits and barriers. Discover what other agencies are doing and learn how you can leverage these findings for your agency's GenAI adoption journey.

?? DOWNLOAD THE REPORT ??? https://buff.ly/4ceSGcK ?

?? READ THE STORY ON MARKETINGDIVE ?? https://buff.ly/3VUxKSt?

?? GenAI Blueprint: A Comprehensive Members-Only Guide to Agency GenAI Implementation?

Whether your agency is just beginning to explore generative AI or already integrating it into your workflows, this 4A’s guide provides valuable insights and practical strategies to support agencies in driving business growth, innovation, and delivering unparalleled value for clients.

?? CHAPTERS 1-3 ARE OUT NOW! ?? ?https://buff.ly/3xBNDnB?

THE PROMISE OF SYNTHETIC DATA

There have been rumblings over the last few months that the big LLM players are running out of publicly available data on which to train their models. Synthetic data - or data generated by AI models rather than humans - is one potential answer to this challenge. This type of data has been shown to create models that perform comparably to those trained on human curated data, and it can dramatically decrease time and costs required for model training. NVIDIA recently released its Nemotron-4 340B model, which can generate large amounts of synthetic data that rival traditional methods for training LLMs. A story from The Information speculates that synthetic data could be part of the reason OpenAI and Anthropic perform better than other models against relatively new "hard prompts" introduced by LMSYS.? As with many emerging technologies that show great promise, synthetic data for training seems to carry significant risks that demand consideration as well. Chief among them:

• Do we risk neglecting critical nuances captured only through human experience?
• Could the use of synthetic data lead to homogenized, less creative outputs?
• Might models trained on AI-produced data inadvertently amplify any bias in the original dataset?

As we embrace these technologies, we must tread carefully, ensuring that efficiency doesn't come at the expense of ethical integrity, reliability and human creativity.?

OTHER INTERESTING AI NEWS FOR AGENCIES:

• The copyright issue heats up in the AI music space, with a lawsuit filed by major record labels. Meanwhile, YouTube is rumored to be courting labels for authorized model training.
• California AI legislation under consideration while Silicon Valley joins the fight.
• Toys R Us teased “the first brand film” created by AI, an impressive showcase of GenAI’s video capabilities. But not everyone was happy about it.
• OpenAI has delayed the Voice Mode demo’d during its Spring Update, citing safety and reliability concerns.
• A new study from talent agency UTA says creatives are increasingly using AI in their work, even while their feelings about it are mixed.

RECENT AI CAMPAIGN EXAMPLES:

• Peacock is using AI to clone the voice of renowned sports announcer Al Michaels for personalized Olympic recaps. This AI-generated voice aims to deliver customized commentary, enhancing the viewer's experience with personalized recaps and insights based on the user’s expressed interests.
• Snickers has launched a new U.K. marketing campaign featuring an AI clone of legendary soccer (or football, if you prefer) coach José Mourinho to provide personalized advice to soccer fans. The campaign uses AI technology to create unique videos responding to fans' submitted "own goal" stories, with Mourinho’s digital twin offering customized advice. Developed in collaboration with Helo and WPP’s T&Pm, this effort highlights the potential of generative AI for brands to engage users by producing highly personalized content at scale.
• Zola has launched "Split the Decisions," an AI tool designed to balance wedding planning tasks between partners. By evaluating preferences and strengths, it assigns tasks to ensure both partners equally contribute. Built on OpenAI’s GPT conversational interface platform, the process aims to help couples divide and conquer to manage the mental load of making over 500 decisions typically required in wedding preparations.

Visit our AI/GENAI Hub for the latest guidance from the 4A's.

Missed previous editions of Noteworthy News? Catch up here:

June 21 - #7

June 14 - #6

June 7 - #5

May 31 - #4

May 24 - # 3

May 17 - #2

May 10 - #1

#AI #GenAI #ArtificialIntelligence #EmergingTechnologies