The process of building a successful AI strategy has changed. In November of 2022, the evolution of artificial intelligence took a dramatic leap forward: OpenAI developed a more evolved, smarter chatbot called ChatGPT. By combining generative AI with large language models and highly functional algorithms, the company developed a chatbot supporting a new level of responsive, intelligent behavior.
The recent advances in artificial intelligence are disruptive and require developing new ways of thinking about what an effective AI strategy looks like.
These “smarter chatbots” are powerful tools that can perform research, provide reasonably good writing using human languages, and even write computer code. Smarter chatbots should be considered when developing an AI strategy. (This new form of artificial intelligence is quickly becoming the cultural standard, and soon will simply be referred to as AI. However, for purposes of this article, this new form of AI will be referred to as “smarter chatbots.”)
Because ChatGPT is extremely efficient and easy to use, it was quickly followed by a series of similar smarter chatbots, all capable of communicating in plain, easily understood written English and other languages.
While the language model called GPT (Generative Pre-trained Transformer) has been around for a while, this new version and its variations have crossed a threshold. Modern AI has become genuinely useful for performing a wide range of creative tasks.
The smarter chatbots can generate business ideas, write a wedding toast, and create visual art. Previous generations of artificial intelligence could technically perform these tasks, but the quality of their creations was so low they could be used only for the most basic, simplistic communications.
Since the release of the smarter chatbots, there have been several examples of people using them to accomplish a variety of tasks, ranging from writing children’s books to creating board games to assisting in diagnosing illnesses. The uses of smarter chatbots are still being explored and have great potential – for example, they have recently been used in structural engineering.
The disruptive influence of ChatGPT and its variations should limit AI strategy and AI investment decisions to a range of three to five years, and significant investments in more traditional AI solutions should be carefully considered.
Smarter Chatbots and Data Governance
Atlan is currently promoting the use of smarter chatbots with Data Governance. The company suggests that smarter chatbots can be used to identify recurring tasks in the Data Governance program and can recommend automated processes to reduce manual labor.
The smarter chatbot can also (theoretically) support these processes:
- Detecting anomalies
- Monitoring Data Quality
- Data discovery with complete context
- Regulatory compliance with automation (GDPR, CCPA, etc.)
Traditional AI vs. Smarter Chatbots
The uses of smarter chatbots have not yet been fully explored, which makes it difficult to develop a long-term AI strategy. For the short term, a combination of traditional artificial intelligence and smarter chatbots seems the most efficient path. Traditional artificial intelligence, however, is trustworthy and supported by reliable algorithms, while smarter chatbots are not.
Smarter chatbots can show bias and can even hallucinate. Generative AI, a major component of smarter chatbots can, on occasion, create its own facts and produce “hallucinations.” This, sadly, makes it less than completely trustworthy. The misinformation sometimes provided can range from embarrassing (claiming Mozart invented music) to potential lawsuits (damaging a person’s reputation with misinformation).
In terms of bias, it would seem all forms of artificial intelligence are vulnerable to biases. They are absorbed from the training data, which is heavily influenced by humans’ conscious or unconscious biases. For example, when prompted to show images of “homeowners,” an AI model might predominantly present images of white males.
At present, smarter chatbots can be used for creative endeavors, but have to be checked for errors and misinformation by a human being with the appropriate background. The use of traditional AI, while less intelligent, is also more trustworthy, and can be used for all basic operations.
An AI Strategy That Combines Traditional AI/ML with Smarter Chatbots
The business goals lay a foundation for developing a modern AI strategy. Many AI strategies are developed with the intention of automating inefficient tasks; however, the AI strategy should be designed to support the business’s overall goals, not simply automation.
Potential AI projects can be prioritized based on the business’s needs, the amount of effort required, and the returns on investments (ROI). A successful AI strategy focuses on using the right solution for the right problems, at the right time. As a result, the business will be able to efficiently provide optimum value to the customer base.
Discovering the priorities of different department managers is an important part of developing a successful AI strategy.
The discovery process involves collecting information about the business that will help in developing an AI strategy. Interviews play a significant role in this process. Different department managers can be asked about their business priorities for this year and the next. Knowing the priorities of each department promotes an understanding of their needs. During the interview, specific problems and potential solutions should also be discussed.
Smarter chatbots can be used for creativity, while traditional AI and ML can be used for routine processes.
The best way to learn about ChatGPT and its alternatives is to start working with and experimenting with them – possibly on a business-related project. Accessing the free form of ChatGPT for learning and experimentation requires registration with OpenAI (or one of its free versions).
After developing a sense of how useful the smarter chatbots can be to your organization (as in making a list of potential uses), building a successful AI strategy becomes easier. Knowing the business’s priorities will help in selecting the most appropriate AI software.
Many businesses have failed to use AI effectively, leaving planned projects incomplete. The main reasons for this are the wrong software or a lack of experience and follow-through. A person will start a project, realize the limitations of the software, and then abandon the project when it delivers only limited results.
In developing an AI strategy, consider the activities and workflows that support the organization: payroll, shipping, security, invoicing. These processes are all standardized, repeatable, and rules-based, and can be streamlined with the use of artificial intelligence. A brief list of commonly used AI and ML processes follows:
- Inventory or demand forecasting
- Automation of repetitive tasks
- Churn prediction in marketing and sales
- Predictive maintenance
Personnel and organizational changes should be considered when building an AI strategy. Very few businesses are prepared for the unexpected changes that take place after installing AI solutions. Suddenly, gaps in certain domains and a need for certain skills present themselves. The question becomes whether to hire new staff or to provide training for current staff. (Another option is hiring contractors or freelancers, as needed, for short-term situations.)
The Steps for Developing a Successful AI Strategy
Costs are an issue, and the disruptive influence of smarter chatbots will impact AI investment decisions. As a consequence, AI strategies should be limited to a range of three to five years. Additionally, determining the return on AI investments should be calculated to determine if the investment will pay for itself within the three-to-five-year period.
It should also be noted that some versions of the ChatGPT model may be a better choice for your specific business than others. They are not all the same, and some have been designed with a specific focus.
Listed below are the steps needed to build an effective AI strategy:
- Define the business’s long-term goals.
- Broadly describe the steps needed to achieve those goals.
- Perform a discovery process focused on each department’s specific priorities and problems.
- Become familiar with ChatGPT and its variations for creative projects.
- Research (which may include interviews with department heads) and select the most appropriate, cost-effective AI and ML software for the business’s automated needs.
- List the organization’s short-term goals for the next three to five years, and the AI software needed to achieve those goals.
- Consider the impact AI software will have on staff whose jobs would be altered/eliminated and whether new staff will need to be hired.
Image used under license from Shutterstock.com