Like a Scout, Be Prepared
“Before starting an AI project, businesses should do an readiness assessment and audit of their data collection and usage,” said Briana Brownell, founder and CEO of PureStrategy.ai. “This audit usually determines what teams are using various data streams, how frequently, what are they using it for, and what decisions are being made as a result.” The process will likely be illuminating in and of itself, Brownell said. For example, it can reveal data silos where companies are either duplicating data collecting or underusing what they are getting.
AI is the point, not a feature:
“Companies building out their AI capabilities should look for solutions that are built around AI and machine learning technologies, not vendors who treat AI as a feature added to a legacy product,” said Doug Bordonaro, chief data evangelist at ThoughtSpot. This is where a lot of companies are going wrong right now, Bordonaro said, because they can distinguish between hype and innovation. “Companies need to look for platforms specifically built to handle all the components needed for successful AI instead of a franken-platform of random technologies bolted together,” Bordonaro said.
Get Your Data Right
Organizations that use their own data to train their own AI systems, or to augment third-party ones, will have some of the best results, said Salvatore Stolfo, a Columbia University professor who specializes in computer science, AI and cybersecurity. “However, this data has to be prepared correctly--reconciled, refined, checked for accuracy and checked for biases,” Stolfo said. Real-world data is messy and will only get more complicated as you bring together data from different sources. “But probably the most difficult problem to solve is defining the high-value use cases for the analyses,” he said. “This will of course require AI expertise, but also business savvy and industry knowledge to get the best value from the high cost of the analytics.”
Start with the Staff You Have
You may need to bring specialized staff on board at some point, but you should start with the staff you have, said Michael Meyer, chief risk officer and chief security officer with MRS BPO. “As with all things technology related, they get easier and cheaper as time goes on,” Meyer said. Companies like Amazon, Microsoft, Google and IBM are working to improve and simplify their AI products in order to make it so that they can be used by staffers without requiring years of training or experience, he said. Companies are increasingly making their artificial intelligence products and training available to the public. Microsoft’s AI School, for example, is available online.
Use Tools and Platforms to Make It Affordable
There’s a shortage of top AI talent. Even if you can snag someone, the cost is steep--as much as $300,000 to $500,000 per year plus benefits and stock options, said Eddie Wang, growth hacker at AppSheet. But that doesn’t mean that smaller firms can’t move forward with integrating AI into their operations where it makes sense. “Given the prohibitive costs of AI talent, forward-thinking companies need to identify tools and platforms that integrate with their existing business processes that can provide some degree of ‘intelligent’ value,” Wang said. “These tools clearly are not a full substitute for a competent AI developer, but they can help to maximize the limited resources a company has to work with.”
Ensure Your AI Helps People Work Smarter, Not Harder
If artificial intelligence is making things more difficult for your clients or users, instead of somehow improving the process or end result, then it’s about hype and not helping. “AI is all about bringing simplicity to your end users,” said Bordonaro. Make sure that if you're bringing AI into your work in some way, it’s about improving workflow or making things somehow work more efficiently, not just about throwing a high-tech label on things. “Ultimately, people won’t use a product unless they believe it delivers on the value it purports to add,” he said.
Remember That You Still Need Humans
“This is more relevant for any firm dealing in high-volume, low-value, highly templated and / or repetitive work--for example, standard conveyancing processes, small claims processes, company formation, and so on,“ Wye said. Artificial intelligence can increasingly handle tasks that were previously taken care of by humans, and there are many cases where that makes sense. But human input is still needed, and there are situations where it doesn’t make sense for a task to be done by AI--and might never. As an organization moves forward, that is important to remember. In the legal profession as well as many others, there are subsets of work that can be replaced by combining AI with clever humans, said Alistair Wye, lead product strategist for iManage.