Quality AI Models Don’t Have to Include Controversial or Copyrighted Content, Right? Right?

NAB

Since the beginning of the generative AI boom, there has been a fight over how large AI models are trained. In one camp sit tech companies such as OpenAI that have claimed it is “impossible” to train AI without hoovering the internet of copyrighted data. And in the other camp are artists who argue that AI companies have taken their IP without consent and compensation.

article here

Adobe’s approach is unusual for a major tech company in that it has tied its future to building generative AI products without scraping copyrighted data from the internet.

In an exclusive interview with MIT Technology Review, Adobe’s AI leaders are adamant this is the only way forward.

At stake is not just the livelihood of creators, they say, but our whole information ecosystem. What they have learned shows that building responsible tech doesn’t have to come at the cost of doing business.

"We worry that the industry, Silicon Valley in particular, does not pause to ask the ‘how’ or the ‘why.’ Just because you can build something doesn’t mean you should build it without consideration of the impact that you’re creating,” says David Wadhwani, president of Adobe’s digital media business.

Adobe says it wants to reap the benefits of generative AI while still compensating people for the work that has gone into training data. It also argues that by not indiscriminately scraping the Web there’s less risk of bias and misinformation being output. Adobe’s model has never seen a picture of Joe Biden or Donald Trump, for example, “and it cannot be coaxed into generating political misinformation.” It has not been trained on any copyrighted material, such as images of Mickey Mouse.

On the other hand, tech companies like OpenAI argue that they are able to build more powerful AI models for the benefit of everyone, including artists. The misuse or mistakes in public GenAI tools can be used to improve the model, they maintain.

While Big Tech is the prevailing model for AI use, are Adobe and other supporters of content verification initiatives winning the argument?