A New AI Era

 

With the advent of Generative AI and OpenAI’s launch of their revolutionary chatbot ChatGPT in November 2022, the world has entered a new era of excitement and awe. These foundation models offer transformative opportunities that promise significant scientific, business, and social advancements.

Generative AI has elevated Artificial Intelligence to new heights in the news and business world. It has sparked discussions about AI’s potential to become a global economic driving force, akin to the impact of computers and the internet. Business leaders are now contemplating how AI solutions and technologies can generate sustainable and measurable growth.

 

A vast majority of business leaders acknowledge that AI is now integrating itself into the corporate world. However, because only a minority of executives truly understand it, only 25% of CEOs believe that AI will significantly impact their company in the next three years. Despite this, the private investment community is optimistic. In the second quarter of 2024 alone, $27 billion was invested in AI startups. This enthusiasm might seem irrational, but AI’s recent advancements affect a wide range of industrial domains, such as drug discovery, driverless vehicles, disease diagnosis, military applications, and the Internet of Things (IoT). Rarely has a technology held such profound potential for shaping the future.

 

Companies that seriously consider the potential of artificial intelligence to solve business challenges or discover new opportunities by implementing AI in a sensible, informed, and purposeful manner will gain a competitive advantage.

 

Machine Learning in business

 

Machine Learning (ML), a subset of AI, is a data-driven, model training methodology. It is immensely powerful but only as effective as the data and planning behind it. ML will not solve problems by itself, alone; it requires a well-thought-out plan and clear expectations. Without a robust and transparent strategy, ML may adversely affect a business and its workforce.

To successfully implement machine learning in a business, consider the following aspects:

 

Problem Definition

• Implementing machine learning (ML) requires a significant investment of time and money and can disrupt normal business operations. Business leaders must be clear about the problems they aim to solve and the expected outcomes. They should consider all aspects of project design, cost, and execution to determine if ML is suitable for their business use case. Data quality and availability are crucial early decision factors; acquiring data can be expensive or restricted by legal constraints, potentially making the project unfeasible or cost-prohibitive. By adopting an unbiased perspective and evaluating all the pros and cons, business leaders can make a well-informed and logical decision.

 

Data Access and Cost

• Data is arguably the most critical asset in a machine learning project. Several parameters must be considered, each impacting the project’s cost and feasibility. While this list is not exhaustive, it highlights some crucial points:
  • Data Quality: How will the data be collected, and from which sources? Are these sources diverse and relevant to the project? Are there redundancies or overlaps that need addressing?
  • Data Volume: Is the volume sufficient to represent features adequately? Conversely, too much data can introduce noise and cause model overfitting.
  • Data Diversity: Bias and fairness are significant concerns today. Ensuring data diversity is essential to produce fair and ethical model outcomes.
  • Data Normalization: Normalize and standardize features while addressing missing values. Annotation is critical for helping the model recognize key parameters.
  • Data Privacy and Security: What are the legal limits regarding the dataset? Can the data be used within these legal boundaries? Is the data secure?
  • Data Updates: Can new data be continuously collected to ensure the model evolves with changes in features and data points?

 

Model Design, Training, and Support

• Creating an end-to-end machine learning pipeline—from data gathering and model design to scaling up and production—requires specific technological functions and roles. This undertaking is far from trivial. Companies new to AI might consider outsourcing the work until AI becomes a core business function and area of expertise. Key roles include data scientists, data engineers, software and DevOps engineers, project managers, UX/UI designers, and domain typically being internal company resources.

 

Change Management

• The implementation of any new business process will impact a company’s employees, or at least some of them. They may need to be trained or reassigned to other tasks or functions. Machine learning is no exception, but it may also induce anxiety among employees due to the fear that machines might replace their jobs, similar to how robots have replaced production operators in manufacturing environments.

 

Long-term Relevance

• Leaders must remember that the model they build today addresses the current business needs within a specific market, legal, and regulatory framework. Over time, new market dynamics, customer needs, regulations, and policies may emerge, impacting the model’s validity and effectiveness. The leader of the AI and machine learning initiative must account for these changes, ensuring the model can adapt to the new market reality and evolving business objectives. This will affect the data collected, the model, the learning protocols, and the expected output.

 

Measuring Success

• Launching a machine learning project is likely to be a significant investment for the organization. Determining the project’s ROI will indicate whether the initiative is a sound investment. As the project progresses, a model performance assessment should be conducted to verify if it produces the desired outcomes. This assessment should include a thorough analysis of the model’s strengths and weaknesses, allowing areas where the model falls short to be addressed and improved. Demonstrating tangible value is crucial for gaining additional investment and support in the future and for showing employees the positive impact of the technology on the business. Lastly, accurately assessing the project’s business value will help identify potential competitive advantages that can be leveraged strategically.