Upskilling and Reskilling in the Age of AI

NEW TECHNICAL SKILLS REQUIRED TO DEVELOP AI SYSTEMS

Developers will need many new skills when creating AI systems. Developers should already understand and use general software development tools. These tools include version control, backup and recovery, data analysis and visualization, rapid prototyping, quality assurance tools, and programming language editors. Python is a dominant language for AI development, but R, Java, and C++ are also valuable, depending on the specific application.

Additional specialized knowledge, required to develop AI applications, includes the following areas:

Machine learning. Machine learning is a type of AI that allows computers to learn from data without explicit programming. Learning algorithms take data as input, analyze it, and then use that analysis to make predictions or decisions. 
Mathematics and statistics. A solid foundation in these areas helps with understanding complex AI algorithms and analyzing their performance.
Security, privacy, and ethical policies. Specifying, implementing, and enforcing these policies, guardrails, and constraints will require careful tailoring of both software and human procedures.
Data labeling. This is the process of assigning labels or tags to datapoints so a machine learning model can understand it. Labeling techniques include manual, automated, and active learning (identifying the datapoints that are the most difficult to label manually).
AI software architecture. This is the overall high-level design and architecture of the AI software system.
Developer skills include identifying and relating major modular components, determining if a module should be embedded within the application or accessed externally on the internet, and determining whether to build or purchase the component.
Data management. Developers must know how to turn raw data (text, audio, graphic, video, sensor data, and unstructured data such as email and advertisements) into information usable by AI systems. This includes capturing, cleaning, organizing, and representing large sets of data for building large language models (LLMs) used by GenAI systems.

NEW SKILLS ARE REQUIRED TO MANAGE THE DEVELOPMENT OF AI SYSTEMS

A systems development manager should have these core project management skills: project planning and execution, resource management, communication and collaboration, problem solving and decision making, leadership, delegation, and empowerment. They should have additional skills for dealing with specific issues related to AI systems development.

These are some of the skills related to nontechnical issues:

Legal. Managers must ensure training datasets and generated outputs comply with copyright laws. Understanding the legal implications surrounding user privacy, intellectual property, and potential liability related to AI outputs is crucial.
Environmental. Managers need to be mindful of the computational power and water consumption required by AI systems. Optimizing models for efficiency and exploring renewable energy solutions are important considerations.
Safety. Managers need to assess potential safety risks associated with the AI product and implement safeguards to mitigate them. This might involve human oversight, safety protocols, and testing for unintended consequences.
Black box problem. This problem is the lack of explainability in some AI systems which don’t reveal the reasoning behind their outputs.
Explainable AI (XAI) is an evolving field to explain how AI software makes decisions. This may involve the use of visualization of the internal workings to provide explanations or simplify the AI software while still maintaining good performance.
Hallucinations. Managers are responsible for implementing methods that detect and mitigate AI hallucinations. This might involve using explainability techniques, rigorous testing, and retraining models with high-quality data.
Privacy. Reasons for sharing data with software from outside your organization may include training the AI system or some of its users, synthesis of business strategies, scientific discovery, or personalize services.
Managers need to understand the reason why data is shared and how they can control it.
Methodology. Managers must understand the iterative nature of AI development and how to adapt project plans subject to time and cost constraints.

By understanding and managing these diverse responsibilities, AI project managers can ensure the successful development, deployment, and responsible use of AI systems.

SUMMARY

Many employees and managers may feel uncomfortable or threatened by AI advances in the workplace. Upskilling and reskilling empower both employees and managers to navigate the changing job market and secure their careers and also benefit businesses by building a future-proof workforce where they can thrive alongside AI systems. AI systems are here today, and additional AI systems are coming. Upskilling and reskilling will ensure that you and your business will be effective in the use of this new technology.