The success of an AI program may depend as much on art and philosophy as it does on data science and machine learning. That's because effectively deploying AI in an organization requires building a comprehensive team that includes people from a variety of backgrounds and skill sets, as well as non-technical roles.
MosheKranc, CTO of Ness Digital Engineering, said, "Any AI initiative requires a combination of IT specialists and industry domain experts.IT specialists understand the machine learning toolkit: which families of algorithms are most likely to solve a particular problem? How can specific algorithms be tweaked to improve the accuracy of the results? And industry domain experts bring domain-specific knowledge: which data sources are available? How dirty is the data? What is the quality of the machine learning algorithm's recommendations? Without input from industry domain experts, IT specialists may not be able to answer these questions."
So the conclusion is that the success of AI really does depend on the team, rather than any individual or role.
Keith Collins, executive vice president and chief information officer at SAS, said, "When building an effective AI team, we need to look to industry experts or superteams, and it's teamwork that wins the day. Diverse disciplines are key to AI success."
Four core types of AI talent
Collins sees four core types of people needed on AI teams:
? People who understand that business processes are critical to building real-world scenarios and valuable outcomes.
? People who understand analytics techniques such as machine learning, statistics, prediction and optimization and use them correctly.
? People who understand where data comes from, what the quality is, and how to maintain security and trust.
? AI architects who understand how to implement analytics through results.
Collins noted that, like other IT leaders and AI experts, these core disciplines or roles can draw from a variety of backgrounds. He cited disciplines such as music, chemistry, and physics as examples.
He said, "These disciplines encourage people to understand scientific processes and thinking in the context of complex interactive systems. They typically specialize in the critical thinking skills needed to build good experiments and apply the results of machine learning."
The value of a diverse AI team
The value of a diverse team is wide-ranging: for example, it can help organizations better deal with AI bias. Solving business problems (including the biggest and toughest ones) is also important, which may be one of the reasons organizations develop an AI strategy in the first place.
Jeff McGehee, Senior Data Scientist and Head of the IoT Practice at Very, said, "It is widely recognized that diversity of opinion is critical to solving all complex problems. Diversity relates to lived experience, and professional backgrounds are an important part of most people's lived experience, which can add dimension to AI projects and provide new perspectives on finding innovative solutions."
McGehee also noted that building AI or other diverse teams requires an active effort on the part of the organization and as part of its recruiting and hiring practices. Organizations will find that achieving diversity may not be a viable team-building strategy.
With this in mind, there is a need to understand the range of experts and roles, including non-technical roles, that are valuable to AI teams.
1. Domain experts
One can think of these roles and people as subject matter experts. Regardless of which term is used, it is important to understand their importance to an organization's AI program.
McGehee says, "Developing AI systems requires a deep understanding of the domain in which the system operates. Experts developing AI systems are rarely experts in the actual domain of the system. Industry domain experts can provide key insights that enable AI systems to perform optimally."
Kranc of Ness noted that these experts can solve enterprise- and strategy-specific problems in their domain.
He said the type of industry domain expert depends on the problem to be solved. Whether the insights needed are in revenue generation and operational efficiency or supply chain management, industry domain experts need to answer these questions:
? Which insights are most valuable?
? Can the data collected about the industry sector be used as the basis for insights?
? Are the insights meaningful?
Some industry domain-specific examples are presented below, but first a look at some of the other key players on the AI team.
2. Data scientist
Dave Costenaro, head of AI research and development at Jane.ai, said this is the first of three key requirements for AI teams working on new build projects. Its sample projects include chat agents, computer vision systems, or prediction engines.
Costenaro said, "Data scientists have a variety of backgrounds - statistics, engineering, computer science, psychology, philosophy, music, etc. - and are often intensely curious, which forces them to dive deep into the system to find and use patterns, such as what they can provide to an AI project, determine what it can do, and train it to to do that."
3. Data engineers
Costenaro said, "Programmers take ideas, models, algorithms from data scientists and bring them to life by normalizing the code, making it run on servers, and successfully talking to the right users, devices, APIs, etc. "
4. Product designers
The end result of the three key requirements also illustrates the value of the AI team's non-technical expertise, Costenaro said.
He said, "Product designers also come from a variety of backgrounds, such as art, design, engineering, management, psychology, philosophy. They create the roadmap for what's needed and useful."
5. Artificial intelligence ethicists and sociologists
Artificial intelligence ethicists and sociologists may play a vital role in some sectors (notably healthcare or government), but it seems likely that they will become increasingly important in a wide range of use cases.
McGehee said, "An important part of an AI system is understanding how it affects people and whether underrepresented groups are treated fairly. If a system has unprecedented accuracy but doesn't have the desired social impact, it's doomed to fail."
6. Lawyers
McGehee said that a separate but related need for legal expertise is also being seen in this emerging field.McGehee said, "The GDPR regulations set a precedent for developing regulations around algorithmic decision-making. As the world becomes more aware of the use of AI in industry, more laws are expected to be introduced. Lawyers who are well versed in this area could be a valuable asset."
Because industry domain experts are so important, as Kranc and McGehee articulate, there is a need to look at specific examples of industry domains, both technical and non-technical. These areas should be part of AI team building, depending on the specific goals and use cases of the organization.
Jane.ai's Costenaro noted that "since AI is often just an enabling layer to augment an existing business use case, team members who have supported that use case in the past are still valuable and essential for the same reasons."
Costenaro provides five examples of roles that may be valuable for AI contributors and explains how existing roles can be adapted and augmented in an AI environment.
7. Executives and Strategists
Costenaro said, "Enterprise executive leadership will need to consider which business models can be automated and improved with AI and weigh new opportunities and risks from teams such as data privacy, human-computer interaction, and more."
8. IT executives
Don't be confused about the value of non-technical roles: an organization's AI strategy won't get very far without IT. Costenaro notes that IT teams need to address questions such as, "If large amounts of data are being accumulated and stored for model training, how will data privacy and security be ensured? In addition, how will it be stored and served quickly and reliably from the server to the customer's device?"
Costenaro added that this will also drive growing demand for DevOps professionals and people with expertise in cloud-native technologies such as containers and orchestration. And IT departments have the opportunity to use AI tools like chatbots to streamline internal services.
9. HR Leaders
Costenaro said, "Similarly, there are many opportunities for HR to become more efficient by using AI tools like chatbots to serve customers."
Additionally, it seems likely that HR will be a key player in evaluating the impact of AI within an organization, not unlike McGehee's inclusion of roles such as ethicists and lawyers.
10. Marketing and Sales Leaders
As Kranc points out, if an organization's AI initiatives are related to revenue generation, then consideration should be given to adding domain expertise from areas such as sales and marketing.
Costenaro also noted that sales and marketing professionals may need to augment their existing skills and processes with technologies such as sales automation tools and robotic process automation (RPA) as part of an AI program.
11. operations specialists
Across the IT department, operations and DevOps professionals have specific domain expertise to implement AI initiatives.Costenaro cited the following questions as examples of where expertise needs to be applied:
? What can be automated and improved?
? If machine learning models are used, how will new data collection processes be created to continually train and improve these models?
? Can off-the-shelf, pre-trained models and/or datasets be accessed from open source repositories to get a huge head start? Are there tasks and use cases that are taken into account by the API services offered by third-party vendors?
While AI can solve some major problems, it will certainly create new challenges. This is the fundamental reason why diverse teams are constituted.
McGehee said. "Having people with different backgrounds and personalities focusing on different project details and constraints is useful because it improves the likelihood of all the important details and provides a holistic approach to identifying solutions."