Recent years have seen a surge of investments by organizations in piloting and deploying artificial intelligence (AI) in a variety of applications. Building effective AI deployment teams has challenged organizations for a variety of reasons, including a demand for college-educated professionals that far exceeds the supply. In addition, organizations face issues with building much-needed diversity in their deployment teams. <\/p>\n
<\/p>\n
<\/p>\n
Partially to address the talent gap, many of the world’s most well-known technology companies have dropped the requirement<\/a> of a college degree in their hiring process.<\/p>\n Recognizing that self-starters have found non-traditional ways to become highly skilled in computer architecture and software development, companies like Apple, Google and IBM have increased hiring of non-degreed practitioners.<\/p>\n College-educated professionals often still command a salary premium, but self-taught computer experts are now eminently employable, because the search is for competency rather than formal training. But this does not mean that literally anyone with a solid foundation in software development and artificial intelligence should be part of an organization's AI deployment efforts. <\/p>\n <\/p>\n Whether college-educated or self -trained, there are a number of important skills that AI deployment team members should have to become truly valuable assets. This article will highlight some of the common competencies necessary in an effective team. (Read also: The Ultimate Guide to Applying AI in Business<\/a>.)<\/strong><\/p>\n AI deployments span a wide range of applications, and each individual deployment can require very specific competencies. However, there are general skills desirable for team members no matter what the deployment and no matter what their educational background. Identifying these foundational skills requires analyzing the AI deployment process.<\/p>\n <\/p>\n Initial stages of the AI lifecycle, include determining the business requirements for use of AI and the analytics approach to be used, identifying and collecting relevant data, and building and validating the model. Deployment is one of the final steps in the AI lifecycle, but deployment team members should be familiar with the entire lifecycle, as well as their place in it, particularly as they may be involved in many phases of the process.<\/p>\n <\/p>\n At a high level, the AI lifecycle is very similar to the software development lifecycle<\/a>, which, according to software engineer Mark Preston of Cloud Defense<\/a>, “defines different stages that are necessary to bring a project from its initial idea or conception all the way to deployment and later maintenance.”<\/p>\n <\/p>\n Given the structure of the AI development lifecycle, the team as a whole should share a common foundational knowledge base that includes:<\/p>\n <\/p>\n A general understanding of the business requirements for AI deployment.<\/p>\n<\/li>\n A general understanding of the analytic techniques to be used (e.g. NLP<\/a>, machine learning, etc.).<\/p>\n<\/li>\n A general understanding of the data to be used in the AI model.<\/p>\n<\/li>\n A general understanding of the desired outcomes.<\/p>\n<\/li>\n<\/ul>\n <\/p>\n The team must be able to deal with large quantities of data to develop relevant models and recognize patterns in the data. A deployment team may include everyone from AI engineers<\/a> to data scientists to coders to ethicists to IT professionals and more, each with their own specialized knowledge and training.<\/p>\n <\/p>\n For many companies, AI engineers and data scientists are the hardest to come by due to the shortage in the market. Interestingly, AI itself is being used to help non-graduates fill gaps in corporate technical staffing. With an understanding of the development framework in mind, several core competencies emerge, many of which can be met without requiring a traditional college degree.<\/p>\n While this list of competencies is directed to the technical members of the team, many apply to all team members (e.g. lawyers, HR, ethicists, ets.). Reviewing the competencies that help define a successful technical member of the deployment team makes clear why a four-year college degree is not always a necessity.<\/p>\n AI deployment requires continuous monitoring and modification, so one of the key competencies for team members is the ability to learn. In fact, AI engineers and data scientists may require as much training as their models because the AI field is rapidly evolving, with new algorithms appearing frequently.<\/p>\n <\/p>\n A college degree is not the only indication<\/a> of an ability to learn in the future. One could argue that a self-taught AI practitioner has demonstrated more of an ability to learn than the college-educated team member, precisely because their education was not guided. Indeed, they first had to learn how to learn. One way that both degreed and non-degreed candidates can show both their technical knowledge and their ability to learn is with current industry certifications.<\/p>\n Naturally, one primary requirement for the AI engineer or data scientist is strong technical knowledge. Technical members of the deployment team need to be familiar with neural networks, deep learning, algorithm development, software development, programming, data science, and statistics, as well as other AI fundamentals.<\/p>\n <\/p>\n Both degreed and non-degreed team members should invest time in understanding non-technical issues. Among other things, the deployment team should understand compliance and risk management issues, including the general ethical and regulatory issues involved in AI deployments, as well as the potential risks associated with the specific deployment.<\/p>\n For instance, team members should understand the potential for bias insertion<\/a> in AI models. By understanding issues such as AI bias, the deployment team can ensure the models are appropriately trained, tested and validated prior to deployment. And a more diverse team is more likely to identify those issues. (Read also: Fairness in Machine Learning: Eliminating Data Bias<\/a>.)<\/strong><\/p>\n The most effective team members will couple their ability to learn with a demonstrated desire to learn. Curiosity and creativity will help team members to keep up to date with the most cutting edge algorithms and training methods, and will help them to better optimize their models after implementation.<\/p>\n Team members also need to consider whether there are data privacy issues involved, as this may affect design and implementation of the AI model. That’s because several of the most popular applications for artificial intelligence involve sensitive assets or data.<\/p>\n <\/p>\n The good news is that there are a number of powerful AI-based tools designed to protect these applications. Specifically, AI security tools work by searching through vast amounts of data and searching for patterns<\/a> that indicate if a security breach has taken place, such as from malware. What’s more, is that AI can be used to detect security breaches in applications that are held outside of traditional computer systems, such as in the internet of things or the cloud.<\/p>\n The diversity of teams require that team members be able to communicate effectively. Particularly where complex technical concepts need to be conveyed to non-technical audiences, both inside and outside the organization, it is important that the AI engineer have well-developed communication skills. (Read also: Why Does Explainable AI Matter Anyway?<\/a>)<\/strong><\/p>\n <\/p>\n For organizations seeking to employ non-degreed practitioners, it is critical to assess their communication skills as part of the hiring process, as they may not have had the same public speaking opportunities typically experienced during college.<\/p>\n Organizations need a wide range of competencies<\/a> to ensure successful AI deployment. Deployment teams must have not only the relevant technical expertise to develop effective models, but also the necessary knowledge of the market to ensure that their models are useful, as well as an understanding of ethical and regulatory issues to ensure that the deployment is defensible. (Read also: Expert Predictions for AI and ML in 2021<\/a>.)<\/strong><\/p>\n Building a team around core competencies, rather than whether every team member has a four-year college degree or better, allows teams to create more diverse and effective deployment teams.<\/p>\n","protected":false},"excerpt":{"rendered":" Recent years have seen a surge of investments by organizations in piloting and deploying artificial intelligence (AI) in a variety of applications. Building effective AI deployment teams has challenged organizations for a variety of reasons, including a demand for college-educated professionals that far exceeds the supply. In addition, organizations face issues with building much-needed diversity […]<\/p>\n","protected":false},"author":7882,"featured_media":50548,"comment_status":"open","ping_status":"open","sticky":false,"template":"","format":"standard","meta":{"_acf_changed":false,"_lmt_disableupdate":"","_lmt_disable":"","om_disable_all_campaigns":false,"footnotes":""},"categories":[573,577],"tags":[],"category_partsoff":[],"class_list":["post-50547","post","type-post","status-publish","format-standard","has-post-thumbnail","hentry","category-artificial-intelligence","category-it-careers"],"acf":[],"yoast_head":"\nAI Deployment Needs in Software Development<\/span><\/h2>\n
\n
Essential Competencies for AI Deployment Team Members<\/span><\/h2>\n
1. Ability to Learn<\/h3>\n
2. Relevant Technical Knowledge<\/h3>\n
3. Curiosity and Creativity<\/h3>\n
4. Understanding of Security and Data Privacy<\/h3>\n
5. Communication skills<\/h3>\n
Conclusion<\/span><\/h2>\n