Artificial intelligence and related technologies are in demand, with business leaders anxious to try them out to see how they can improve their visibility, analysis, and forecasting. Generative AI, of course, is democratized AI, and is now easily accessible to everyone. However, behind-the-app AI -- the kind that is embedded into systems and is poised to deliver the earliest tangible value to business -- is not so accessible, and more difficult for businesspeople to grasp.
Such is the tone of a recent Twitter essay by Rachel Woods, research data scientist formerly with Meta/Facebook, who cautions that while solid business cases may be forming around AI, usability is still out of reach.
"AI still has a major usability problem," she wrote. "Most people are still grappling with how to leverage tools like ChatGPT/LLMs/Generative AI. Everyone's waiting for someone to tell them some secret killer use case. We're left with a lot of people questioning the practicality. But these clickbait articles fail to name the underlying issue: It's not that these tools are 'not useful'... They actually just have major usability problem "
Other industry observers agree, at least to a large extent. "The reason why ChatGPT, and overall AI, caught on fire is because of the ease of use, and exploration of the art of the possibility by business users in very simple terms," says Andy Thurai, principal analyst with Constellation Research, points out. "Particularly generative AI's ability to produce text, content, video, and audio, which blew away non-tech savvy users on the potential of AI."
Technology professionals, on the other hand, "who have been limiting their usage to non-tech users due to various reasons such as bias, technology limitation, liability issues, and such, were blown away by the overwhelming response and immediate adoption," Thurai says. "This gave confidence to the creators of the adaptability, but also it took away the need to explain things."
Still, for the most part, only "a relatively small number of people" have a deep understanding of AI, says Dr. Vishal Sikka, founder and CEO of Vianai. He puts the number at about 20,000 to 30,000 people around the world. While there are about a million data scientists in the world, "many of them could not tell you why the system is doing what it is, why it makes the recommendations it does, what could possibly run awry, or how the underlying techniques work," Sikka states.
There is a divide between enterprise use cases for AI and generative AI, requiring different use cases and approaches. "Just producing content is not enough," says Thurai. "It needs to solve a business problem. It should be responsible, ethical, explainable, and auditable, and should be defensible on originality and decisions made. Those are more than just usability issues. These issues can bring down any enterprise."
Enterprise adoption will be slow, but use cases are coming to the fore. "From what I have seen, legal, HR, ethical, and finance teams are all involved in exploring use cases that will bring a lot of value to them," says Thurai. "AI can be expensive, especially if done the wrong way. It can cost them their existence, so care and caution needs to be exercised before jumping in with two feet in this gold rush."
ChatGPT made AI incredibly accessible, but "there's still a significant ramp up period to where the real value is," Woods adds. "To find your breakthrough use cases, you're going to either have to put in the work, or wait until it's so mainstream and the usability is more solved."
What should technology proponents do to increase the usability of AI? Industry experts offer some ways to get started:
Frank, open communication on the possibilities and challenges of AI: A skill that technology managers and professionals need to develop more thoroughly is the ability to sell the right approaches to AI to their businesses. "One of the reasons why we see many technology or innovation ideas fail is because it fails to get traction from business users and from budget holders and CXOs, if they fail to see the value it brings to their enterprise," Thurai says. "Granted the other way is true as well. Techies shooting down the business user needs as impossible to execute or due to budget, technology, resource, cost limitations."
AI education for all: "Companies of all sizes need to increase their tech literacy enterprise-wide to create a wider range of talent working on the AI systems in place," says Sikka. "More employees must be educated on the transcendent aspects of AI in particular. They need to learn the limitations and the weaknesses. Not just what it can do, but things that it cannot do and what needs to be built in an AI system that compensates for these limitations."
Collaborative workshops: Thurai advocates for the use of "collaborative workshops where we invite techies or implementers, innovators, strategists, evangelists, business users, budget holders and CXOs to explore joint use cases. Once they see the proposed use cases in action their minds open up. They start exploring potential use cases that add value to them."
Build your AI talent pool: AI is and will remain a narrow skill as are many other areas of technology. "Photoshop, Excel, Facebook Ads Manager -- all are skills," says Woods. "It probably takes 100-plus hours to hit the tipping point where it becomes second nature to integrate it into your daily work and life."
Ultimately, AI needs to be human-centered, Sikka advocates. "Too many systems aren't designed for humans," he says. "We need to bring the power of human understanding together with data and AI technology -- human-centered AI. This can create intelligent systems that will greatly improve business outcomes and processes as the feedback from humans will naturally improve the AI's performance and outputs."