Well, that was quick. The hype over last week’s new AI demos by OpenAI and Google has already metamorphosed into fears of deepfakes and IP ownership. The claims from Scarlett Johansson’s lawyers are big tabloid fare. They claim that, at best, OpenAI mimicked her voice for their new AI assistant, and, at worst, they trained their models using her voice. Although OpenAI removed the ScarJo-esque voice, you can judge the similarities for yourself with this video. (I see the inspiration, but I’m not convinced this is infringement.)
While this plays out on TMZ, Page 6, and SNL, the scenario will quickly move to questions broadcasters will have to answer. Will Al Roker or the local TV meteorologist give a hyperlocal weather forecast on a streaming service? What about a big-city morning DJ reporting the suburban headlines or narrating your kids’ lunch menu on Alexa? The technology to do this is already available for the broadcasters that want to connect the dots. The challenge is with the IP, consumer demand, and ability to find a business model that works.
AI Headlines to Know
Major AI Companies Agree on Creating an AI ‘Kill Switch’
Google Adds AI Summaries to Search Results — Complete With Ads
Microsoft Highlights CoPilot+ Laptops With AI Built-In
Apple to Add OpenAI to Add Chatbot to iOS18
GenYs/Zs Say They Aren’t Getting Enough AI Training
According to Deloitte’s State of GenAI in the Enterprise survey, 37% of business leaders reported that their organizations are slightly or not-at-all prepared to address talent concerns related to GenAI adoption. Furthermore, less than half believe they are adequately educating their employees on the capabilities, benefits, and value of AI. This sentiment is echoed among Gen Z and millennials, with only 51% and 45% stating that their employers are providing sufficient training on GenAI. Read more
Amazon Plans New Alexa Subscription Service
Amazon won’t comment yet, but plans have leaked for a new AI-powered subscription service for Alexa that will cost above-and-beyond the cost of Prime membership. Over the last decade Amazon has failed to turn the smart speaker into a profit center, and CEO Andrew Jassey is rumored to want to inject an AI subscription model into Alexa to jolt Amazon back to the forefront of the AI discussion and make the 500-million smart speakers in the market more responsive to users’ needs. Read more
ELECTIONS AI IN THE UNITED STATES:
Yesterday, FCC Chairwoman Jessica Rosenworcel proposed that the agency should require disclosures when AI-generated content appears in political ads on radio and TV. If adopted, this proposal would initiate a proceeding to ensure consumers are informed when AI tools are used in political ads, with the FCC seeking comments on specific disclosure requirements. The proposal includes on-air and written disclosures for AI-generated content in candidate and issue advertisements.
Meanwhile, the Senate Committee on Rules Administration reviewed two AI bills affecting media companies’ election content and advertising.
- The first bill, the Protect Elections from Deceptive AI Act, makes it illegal to knowingly distribute deceptive AI-generated election-related videos or images, unless news stations use them with a warning or if the content is parody.
- The second bill, the AI Transparency in Elections Act of 2024, says political ads must have a clear label if they use AI-created content.
Although both bills have been voted out of committee, they still need to be passed by both chambers. Republicans currently object that vague language may inhibit free speech rights.
Meanwhile, attorney David Oxenford reports that eleven states now have laws in place regulating AI in political advertising, some potentially placing liability on media companies who distribute the deepfakes.
MEANWHILE IN INDIA:
This year’s general election in India continues through June 1 and has inundated the country’s one billion voters with AI-generated advertising and deepfakes. Sanctioned AI-messaging has quickly become a $60-million dollar industry. Some candidates are using AI to provide personal outreach at a massive scale, including translating a candidate’s single message into all of the 22 languages and thousands of regional dialects spoken on the sub-continent. These AI-generated phone calls and videos are allowing politicians to reach remote areas where they cannot visit physically.However, there have been legal issues and bad actors. The Election Commission has issued warnings, and there have been police complaints regarding unauthorized AI-generated endorsements. The potential for misinformation is high, and tracking down the original content to verify authenticity can be challenging.
Related
China Is Using AI News Anchors to Deliver Its Propoganda
Below is a prompt for ChatGPT that shows how much the technology has come in only a few months. This prompt wouldn’t work using AI models three months ago.
In February, when I attempted to use this prompt which prospects for advertising clients using ChatGPT 3.5, the results at first looked amazing. “Look at all the new businesses in my suburb I had never heard of!” Checking the AI‘s work proved everything was a (well-crafted) hallucination.
Moving forward to ChatGPT 4, the coders patched the hallucinations with a response that the AI could not provide the real-time information and gave no information beyond that.
Using the latest version — ChatGPT 4o — the results look pretty good. Try them out for yourself; but with this stage of AI, it’s still trust, but verify.
Instructions: Replace the XXXXX with the variables you want to search for. Then paste everything between the asterisks into ChatGPT 4o, which is now the free, but not default, AI engine all OpenAI accounts at http://chatgpt.com
******************************
CITY = XXXXX
RADIUS = XXXXX
ADVERTISING MINIMUM = XXXXX
REVENUE MAXIMUM = XXXXX
You are a marketing professional who helps SMBs increase their business in a [RADIUS] mile radius of [CITY]. Find 10 businesses who can use help with advertising strategies that can afford spending [ADVERTISING MINIMUM] per month on advertising. List the business name, phone number, and a contact person. Do not include businesses with annual revenue over [REVENUE MAXIMUM] and do not include chambers of commerce. Review all results to ensure that you are not hallucinating and the businesses, phone numbers, and names exist in real life.
Race with AI-Powered Cars Quickly Goes Off-Track
The inaugural race of the Abu Dhabi Autonomous Racing League (A2RL) at Yas Marina Abu Dhabi Grand Prix Formula 1 track was a spectacle of high-tech mayhem. During qualifying trials, AI-driven race cars randomly juked, spun out, crashed into walls, and even took unscheduled breaks off-track. All the while, the patient announcers maintained their composure, lauding the event as a push into the future of racing.
When the actual race began, chaos ensued once again. The leading car spun out on the fourth lap, causing a yellow flag. True to their programming, the AI other cars behind obediently halted, unwilling to break the rules and pass during the caution. After an hour, the eight-lap race concluded. Here’s to hoping for smoother laps in the future! Read more
Originally published by Jacobs Media
