Why the Power of Gen AI Lies in the Augmentation not Automation of work. And how Decision Intelligence can maximise the effectiveness of augmentation.
Much of the hype surrounding Gen AI has been related to its perceived potential for replacing human workers with AI software agents with all the implications that has for the job market and for the dangers of rogue autonomous agents.
Beneath all the market hype, Gen AI is not well suited for autonomous enterprise operations due to its inability to automate reasoning & planning tasks in a consistent, predictable, explainable, and hallucination free manner. This is particularly problematic in automation applications requiring automated decisions and actions in industries where regulations, safety and risk mitigations are critical.
On the other hand, Gen AI can be a transformative technology when it comes to work augmentation possibilities. Don’t think of Gen AI as replacing workers but think of it as making them better / multi skilled by providing them with on tap proactive advice and recommendations as they work! Examples of this are empowering case workers and customer service and technical support staff with best practice enterprise knowledge and guidelines.
Decision Intelligence can deliver Expert Augmentation
The strengths of Gen AI / LLMs stem from the ability to generate relevant content / answers in response to natural language prompts and the ability to extract nuanced answers from such content. However, this puts the onus on the human worker to ask the right questions (prompts) consistently in order to receive the right recommendation / guidance to enable them to make decisions. Furthermore, the human worker also needs to determine what additional enterprise data is required to help make the decisions. Imagine the productivity gains if we can automate the expertise required to ask Gen AI the right set of questions in a structured and consistent manner, and to retrieve additional enterprise data as required, and to finally streamline and present recommendations to the human workers. This is what we can achieve by combining LLMs with Decision Intelligence in what we call Expert Augmentation.
Applications of Expert Augmentation include risk assessment, product recommendation, problem resolution, next best action and email processing. In such applications, the Decision Intelligence engine builds an Expert Augmentation system by
- Constructing a decision map of the sub decisions / recommendations that need to be made.
- Defining the (structured data) attributes that feed these decisions and the associated decision logic.
- Determine whether these attributes are available in structured data sources, can be extracted as nuanced answers from documents / emails using LLMs, or if the human worker is to be prompted for these values.
- Using the captured attributes to infer decisions / recommendations.
- Displaying the decision / recommendations to the human worker to make the final decision based on structured explainable recommendations and the attributes driving these.
Combining Gen AI / LLMs and Decision Intelligence enables enterprises to combine the strengths of both technologies to:
- Use enterprise structured and unstructured data for making recommendations / decisions.
- Combining symbolic and predictive AI reasoning with Gen AI reasoning
- Applying rail guards, constraints, and verification to Gen AI reasoning
- empowering the human in the loop with best recommendations based on insights from structured / unstructured data and best practice enterprise expertise
Applications of Expert Augmentation include:
- Processing incoming communications, loan applications, insurance claims etc.
- Assessing operational performance / risks / readiness and recommending actions
- Supporting complex customer service conversations
Follow XpertRule on LinkedIn to get a first look at our Gen AI series, including upcoming pieces on each of the key applications for Expert Augmentation.
Or Get in touch to find out more about how Decision Intelligence can make all the difference in your Gen AI adoption journey,
