As a society, we’re still adjusting to the discoveries that we’ve made with these large language models that are so capable that they are creating their own stuff – albeit, taking human material and creating new original products and results.
Some of the biggest questions in applications of AI revolve around the ownership of data – creative data, personal data, data based on our movements, or our expressions. Data showing how a human would perform a job. Data revealing a person’s likes and dislikes. Etc.
You could go on and on with this. But here are three of the big issues that I’m hearing talked about at conferences and events where we discuss the impact of AI in our lives.
AI and Intellectual Property Concerns
First, there’s the question of AI outputs – how large language models work, and whether they are considered fair use of the original human intellectual property. For example, if you train an AI on Michael Jackson’s dancing or singing, how much is the result tied to whoever owns those rights?
Experts who are looking into thus talk about web scraping and the legality of related processes, as well as how we safeguard intellectual property on the web. In some ways, this is a conversation that’s been going on since the beginning of the Internet, where it’s always been easy to copy and paste a posted image that may contain something trademarked or copyrighted.
“AI chatbots like Google Bard and ChatGPT rely on continuous input from information gathered through web scraping and data scraping,” writes Sorab Ghaswalla in a LinkedIn post covering the phenomenon. “Web scraping involves extracting information from websites, while data scraping focuses on gathering specific pieces of information. This process is considered a necessary evil, as it informs LLMs and enables them to generate increasingly accurate and relevant outputs. Without this data, the development of gen-AI, one of the most significant technological advancements, might be hindered.”
That itself has spawned a raft of lawsuits, but now you can easily re-create content based off of intellectual property in any format – image, video, performance, etc.
So we’re going to have to work these things out in court, and as you can see, it can be sort of convoluted.
This question has to do with the models that companies use for their AI applications.
Should a company build its own AI entity from scratch, or use a model or engine supplied by a vendor?
This is a big question in enterprise, and a lot of criteria apply to the answer. It may have to do with the company’s size and specific needs, or a need for privacy and creative control. It may have to do with what leads to faster time to market or other parts of an executive strategy.
We posted a piece from Davos this past January, where Zurich Insurance CTO Archana Vohra Jain talked about using vendor products to get to production phases faster.
“We want to make sure that we are able to make a difference, and go to market quickly,” she said. “There is a lot to be done apart from just building the model.”
Siva Ganesan, Global Head of AI Cloud Offerings at Tata Consulting Services, talked about how AI has changed, and the range of opportunities, as we noted in the past coverage:
“Five years ago, ten years ago, conversations would be: ‘how do you define schemas, how you model logical data structures and the like?’ – today, generative AI has opened the door to say: ‘it doesn’t matter, structured or unstructured, let me complement what you already have in terms of structured ways of interrogating data – let me build the probabilistic scenario for you’ … and I think that’s a new paradigm.”
So there’s more choice for business now, but the leadership still has to make those fundamental decisions, looking at any relevant trade-offs.
User Content Ownership in AI and Digital Ecosystems
This third one here is pretty important, too, and it’s basically consumer-facing, tied to an individual’s rights as a person
It’s also been an issue in the past, but is supercharged with the power of AI and large language models.
Take the European GDPR, the general data protection rule initiated years ago in the Eurozone. What was its purpose? The GDPR is used to safeguard sensitive and personal user data about someone who lives in a European Union country.
Importantly, the state of California has devised its own similar regulation, although America doesn’t have one as a whole.
The basic idea is to keep other parties from exploiting personal data. As we talk about how to use AI, there’s sort of a consensus that personal data should be owned by the person to whom it is attached, or from whom it was derived. Will.I.am, at a recent event, famously supported this idea as he talked about how we’ll use AI practically in the future, and that’s a very interesting interview discussion…!
We’ll have to think about these questions as we move forward with artificial intelligence in nearly every part of our lives.
