Demystifying and Democratizing Artificial Intelligence
There’s been a lot of discussion about what Artificial Intelligence (AI) is, and what it isn’t. AI has been portrayed as the harbinger of the end of the world, a job-killing automaton, or the moment humanity pivots to a singularity. In today’s world, AI is poised to become humanity’s mundanity remover, taking repetitive non-essential tasks and making them one-click assistants for citizens, consumers, and knowledge workers. AI can take a step beyond eliminating the routine by augmenting creativity in humans and enabling them to meet new demands faster – to create more content, create better content, and create content faster. For example, for the creative people who use Adobe’s products, graphic designers can use AI-assisted search to find stock images on criteria like emotions (search for “happy”) or abstract concepts (search for “jump” or “love”). Filmmakers can review footage and have edits suggested to them. And other creative professionals can have their tools like Photoshop personalized and customized to their skill level and work focus.
But today, because of the explosion in the amount of information available, and through advances in and access to technology at large, AI is no longer just the domain of the most advanced organizations in the world. Instead, this technology has been democratized to impact the ways millions of people live and work. This progress has both positive and negative implications which should be openly debated and addressed.
Adobe is excited about this important conversation. Government leaders around the world, and in Congress and the Administration here in the United States, are now seeking to advance the understanding of AI and the way it is developed and used. For example, the House Resolution recently introduced by Representatives Brenda Lawrence and Ro Khanna is an important step forward as we discuss the transformative value of AI and the potential measures we may need to take to enhance transparency, fairness and accountability. In the other chamber of Congress, Senators Rob Portman and Martin Heinrich are launching the Bipartisan Senate AI Caucus. Likewise, the Administration’s recent executive order acknowledges the importance of advancing America’s leadership in AI and ensuring its responsible development and use.
As individuals increase their use of AI, and as more people have the ability leverage it, there should be a heightened discussion surrounding open data sets, ethics, bias, job training, IP protection and data protection. All of these areas are important ones to address early and often with the ever-expanding uses of AI systems.
On the specific topic of bias, poorly-trained AI models and the datasets used to train those models can reflect biases. That’s why the process to minimize bias in outcomes starts with the people involved in the creation of these new technologies and frameworks. An example of this is if you use a dataset to train an AI lending application that only has male successful outcomes, the AI will necessarily exclude women from getting loans. That’s a real problem, and one that is a fault of the engineering practice used to train the AI. The onus is on technology companies to take the steps necessary to institute bias detection in developing and deploying AI systems.
At Adobe, our mission is to serve the creator and respect the consumer. In the area of AI, this means working with datasets that represent a global society; creating systems that don’t amplify stereotypes; auditing and testing our findings to verify that we avoided bias; communicating our processes; and doing everything we can to ensure our teams have diverse and varied backgrounds.
The principles that guide our current AI work serve as the foundation for our AI development and are fundamental to how we approach new areas of artificial intelligence. We welcome the ongoing conversation between government, academia, consumers and industry leaders to shape and mold how we develop and use this critical technology.