The Five Key Takeaways of This Blog Post
- OpenAI is facing legal troubles after unveiling an A.I. voice assistant that sounds eerily like Scarlett Johansson. The suer is none other than Scarlett Johansson herself (and her legal representation, of course).
- This follows more lawsuits plaguing the A.I. industry concerned with the use of copyrighted works in training generative A.I. platforms. Is it fair use, or unauthorized use? is the question that courts of law must decide here.
- This is a cautionary tale for business owners who plan to integrate customizable build-an-A.I. platforms into their operations. Name, Image, Likeness and “persona” infractions can put you in hot water.
- Further issues for business owners may be the problem of plagiarism when using A.I. that is trained on copyrighted works. A business owner must be wary of this when training its own A.I. platform on chosen data.
- Overall, it will be wise in the future for CTO’s looking to integrate and customize A.I. to seek legal counsel to guide that process.
The Backstory
OpenAI CEO Sam Altman approached Scarlett Johansson twice, asking if she had interest using her voice for an A.I. project.
Her answer was no, both times.
Why her, you may ask?
High celebrity profile aside, she provided the voice of an A.I. conversational computing platform in the Spike Jonze movie “Her”.
Johansson’s performance in that film is perhaps the most famous A.I. voice in fiction, besides maybe the voice of the HAL 9000 in Stanley Kubrick’s 2001: A Space Odyssey.
So, Sam Altman approached her (pun intended, and somewhat inevitable) about this famous movie star about becoming the voice of an A.I. voice assistant that OpenAI had in the works. It is among the brazen and extravagant business moves you make when Microsoft pours billions of dollars into your startup.
Johansson rejected the offer of becoming the official voice of potentially the biggest A.I. voice assistant since Siri and Alexa.
So, what did OpenAI do? Well, it went ahead and made a chatbot that sounded quite close to…her.
Okay, to be fair, the company claims that it was not its intention to create a chatbot that imitated Scarlett Johansson. But at the same time, it cast for the “Sky” voice (there are multiple voice options for the voice assistant) a voice actress who happens to sound just so very close to the famous actress who rejected OpenAI’s offer to voice the A.I.
Where Things Are Now
So, naturally, this led to a lawsuit by Scarlett Johansson against OpenAI.
As a result, OpenAI put the release of the Sky voice on hold until this whole thing gets worked out.
Until then, there is not much to say beyond speculations about how this will work out. Considering these legal troubles, business owners can learn important lessons for responsibly using A.I. in their own operations.
The Significance to Business Owners
One of the more attractive aspects of A.I. to business owners is its customizability.
Businesses can give A.I. data specific to their operations to integrate the A.I. more easily.
This can lead to creating custom chatbots and even voice assistants down the line.
So, okay, your business, even if it is an enterprise-sized operation, may not have the cash at hand to approach famous movie stars to sell their voices to your customized A.I. project.
But be warned, doing what OpenAI did, in creating an A.I. that may have a strong vocal resemblance to a famous person or just well-known voice could land a company in profit-harming legal troubles.
The temptation to use an A.I. that sounds like anybody or anything (imagine a company having their customer service voice A.I. assistant sound close to Siri) is likely strong for some business owners, largely because the familiarity of the voice could lead to positive associations with the business for consumers.
This raises concerns about what data is legally permissible for training A.I. chatbots. Numerous lawsuits claim ChatGPT used copyrighted works without consulting or paying creators.
The result, then, is that business owners will need to be disciplined in choosing what data to train its custom A.I. models on, as well as what characteristics those models will have.
Recent Comments