One of the greatest barriers to artificial intelligence (AI) adoption is trust. Amongst other things, this includes the defensibility of results, leadership’s support in both the sourcing and roll-out of AI solutions, and comprehensive roadmaps for continued professional development. All these examples seek to remove the mystery of AI. Mystery in this case refers to the opaque nature of how AI tools are rolled-out, and not a basic understanding of AI via its academic or practical definitions and general purpose. In the title of this post, I deliberately use the word ‘around’ instead of ‘about’.

When we do not fully understand what something is, we might need to learn about it. If we already understand what AI is, then our growth comes from studying around it by navigating its application specific to given roles across an industry or profession. I might understand what machine learning (ML) is in a general sense. I may be able to communicate the concept of ML and use that referred knowledge whenever the subject arises. But there is much more to be done if I hope to truly understand what its applications are as it relates to specific roles. Broadly approaching a topic like ML creates the risk of opening Pandora’s box.

The process of successfully adopting AI at scale is simply to involved to not have some form of industry certification. There is enormous value in the collaboration of industry professionals sharing ideas to navigate common problems. I have seen this every week on The Cowen Café; a weekly think-tank run by David Cowen of The Cowen Group. Like clockwork, 60+ leaders representing some of the largest law firms and corporations in America (and increasingly Europe) log-in to discuss challenging ideas and share experiences. The value this provides cannot be understated. Trying to navigate the implementation of technologies like AI without outside assistance is a herculean task. There is not enough time in the day to successful perform one’s job whilst simultaneously tackling the sourcing of an appropriate solution, growing internal support for that solution, and finally deploying the solution without aging at an unnecessary rate.

Steps to Developing Industry-Relevant AI Certification

Regardless of the topic, the same basic steps go into the development of a quality certification program. First, a complete due diligence process is required, defining the program’s objectives. This includes clearly understanding the target audience, the pain points the program is trying to fix, and ensuring there is a national-body plan on launching the engagement. These are just some of the considerations to be addressed in the due diligence process. This is also the time when thought leaders can define the need and ability to support a certification program. It is important to keep in mind that a certification program is not always the right response to a training need or pain.

After due diligence is complete and the organizing body has determined that AI certification is suitable for the legal industry, the following steps are required:

1. Select a product. The next steps in creating an AI certification program are determining what product or service the industry body will design its certification around. It would be unwise to try and build an AI certification program covering all the technologies within the field. It may make more sense to select one or two pathways, such as robotic process automation (RPA) and ML as part of a new educational pathway, while defining alternate ways to support and share knowledge on related AI subfields.
2. Determining program features. This may seem obvious, but it is an important part of the AI certification program development process. Program features are simply “what” and “how” a program operates. Program benefits are the “why” or “when.” In other words, a more technical AI certification will focus solely on AI algorithms and applications, while a program geared for the strategy side will also include the technologies value producing benefits.
3. Develop a curriculum. Keep in mind that even the best trainers do not always make good instructional designers. Training is more of an art whereas instructional design is a science. Designers must have expertise in adult learning theory and are often certified themselves. Curriculum development involves defining the learning objectives, creating content storyboards, sifting through a wealth of information to define content in just the right way, and creating student takeaways.
4. The process then takes an even more difficult turn with creating a test bank of questions to support the curriculum. It is essential that no test question be asked that is not taught in the course content and likewise no course content is included that does not tie directly to a learning objective. Understanding the finer points of writing quality test questions (such as avoiding true/false questions and rarely using “all of the above” as potential answer) is perhaps better left up to the experts such as psychometricians.
5. Next you are ready to test all of the material and test questions for validity and reliability. Working with a strong bank of beta testers will help facilitate this process. It is impossible to truly perform quality control on your own work, so it is also impossible to develop a certification program in a vacuum.
6. Finally, you are ready for program launch and implementation.  Timing is everything when it comes to implementing a program.
7. The last and arguably most important step in certification creation is program maintenance. A program is only as effective as its regular maintenance. In the legaltech industry, technology changes fast as do product features and software builds. It is essential that an AI certification program keep pace with industry by releasing regular updates.

Certification is an ambitious goal, so it is important to consider several questions in the process of developing a suitable AI education program. This includes understanding whether the target audience has already achieved an advanced educational degree or employment-level training required to accomplish the requirements of their professional position. It also requires peer consensus on why the certification is ultimately necessary. Further, successful certification cannot be a flippant process to attract interest. It must be relevant and rigorous enough that it will lead to employment advancement or increased compensation. Avoiding the “nice to have” and creating an industry recognized “must-have” is critical.

To be successful, certification programs require the direct support of upper management. Developing an industry-focused AI certification program requires time, money, and a lot of patience. Even still, initial development times are dwarfed by editing and launching requirements, even before accounting for marketing. Without a robust marketing engine that includes incentive for certification, you are setting yourself up for failure. For AI to achieve its full potential, industry professionals need more to ensure a common language and access to reliable roadmaps for achieving their specific goals. This involves the development of recognized accreditation pathways like those provided by ACEDS and other professional bodies.