#421 – HOW TO SUPERCHARGE YOUR RISK MANAGEMENT WITH AI PART 1 – PATRICK OW

ChatGPT, whether you like it or not, is an AI-powered tool that has already moved from science-fiction to real-world application at a very rapid pace. It is one of the many AI tools we use daily, like Siri or Google Assistant, where its conversational fluency helps bring digital powers to everyday users.

The good news is that AI and robots will help people to get more done and have free time for more creative, strategic work. AI-powered tools can analyse vast amounts of data quickly and provide insights that enable more informed decision-making, taking over boring and routine tasks and doing them quicker and more accurately than people.

Several big players like Microsoft and Adobe are already using the power of large language models to allow chat-style interaction with their software.

Checkr surveyed 3,000 employed Americans — an equal number of Boomers, Gen Xs, Millennials, and Gen Zs — and found that 85% of American workers have used AI tools to perform tasks at work. Millennials lead the group with 89% saying they’ve used AI at work.

A lesson on writing prompts for generative AI

Like any tool, the usage will depend on how the tool is used.

Humans, with all their biases, must input the appropriate text-based commands, or prompts, into ChatGPT for it to understand and respond to. It’s basically describing what you want the generative AI to do for you and how.

A prompt may contain any combination of questions, statements, definitions, instructions and input data (or context).

Before you get started with AI prompts, here are a few things to keep in mind:

  • Critically evaluate the output – AI performs three core functions: processing data, identifying patterns, and making predictions. It was NOT designed to fact-check its own outputs, which is why it’s important to do your due diligence and manually verify the accuracy and completeness after the response has been generated. Remember, it is only a tool. Exercise critical judgment and validation always.
  • AI model training biases – The data used to train and fine-tune AI models like ChatGPT can contain biases. While efforts are made to minimise biases during the training process, it can be challenging to eliminate them.
  • Human biases – Apart from AI model training biases, bias can also arise from the way users interact with the model using prompts or how prompts are structured.
  • There’s no magic prompt – The Internet is flooded with AI prompts, but it might take some trial and error before you figure out which type is most effective for your particular use case.
  • Clarify your desired outcome – The response will generally be as broad or specific as its prompt. In some cases, it might actually be preferable to skimp on the details; like if you’re brainstorming blog post topics and want more varied and diverse answers. Something more complex, like a sales playbook or chatbot script, will almost always require more details.
  • Garbage in, garbage out – The quality of the output depends entirely on the input. If you do not take the time to flesh out your prompts, you find yourself drowning in responses that are unclear, inaccurate, irrelevant or just wildly off-base.
  • Always be testing and improving – Prompt iteration – tweaking, testing, and refining different types of instructions – will help you generate more usable responses over time. Often, it’s just a matter of rephrasing your prompts, playing around with synonyms or even generating the responses several times.
  • More contextual information required – Contextual information provides a clearer understanding of the specific situation, industry, or organisation in question. This enables ChatGPT to generate more accurate and relevant responses that are tailored to your unique circumstances or requirements, and better aligned with your specific needs, allowing for more actionable insights. With more context, ChatGPT can avoid making assumptions or providing generic responses.

Pros and cons of using ChatGPT for risk management

Using ChatGPT (or similar AI language models) can offer several benefits but also comes with certain considerations.

Pros of using ChatGPT

  1. Access to information – ChatGPT can provide quick access to a wide range of information and knowledge on various risk management topics. It can assist in answering questions, explaining concepts, and providing insights based on its training data.
  2. Idea generation – ChatGPT can help generate ideas and suggestions for risk identification, assessment, and mitigation strategies for further consideration and discussion. It can provide alternative perspectives and considerations that may not have been initially explored.
  3. Risk scenario analysis – ChatGPT can assist in analysing hypothetical risk scenarios by providing insights based on historical data or general risk management principles. It can help evaluate potential impacts, likelihoods, and possible mitigation approaches.
  4. Continuous learning – ChatGPT can learn and adapt based on user interactions. As you engage with the model and provide feedback, it has the potential to improve its responses over time, allowing for a more personalised and tailored experience.

Cons of using ChatGPT

  1. Lack of context and specificity – ChatGPT may not fully understand the specific context, nuances, or details of your organisation or industry. Its responses are based on patterns observed in the training data, which may not always align perfectly with your unique circumstances.
  2. Potential bias – AI language models like ChatGPT are trained on vast amounts of data, which can inadvertently include biases present in the training data. This can lead to biased or inaccurate responses, particularly on sensitive or controversial topics. It is important to critically evaluate and verify the information provided by ChatGPT.
  3. Limited understanding of current events – ChatGPT’s knowledge is based on data available up until September 2021. It may not have access to the most up-to-date information, industry trends, or regulatory changes that have occurred since its training data cut-off.
  4. Ethical and legal considerations – When using AI models for risk management, it is crucial to consider ethical and legal implications. Ensure compliance with data privacy regulations, maintain data security, and address any potential ethical concerns related to AI usage.
  5. Overreliance on AI – Relying solely on ChatGPT or any AI system without human judgment and expertise may not be ideal. It is important to balance AI-driven insights with human judgment, experience, and critical thinking.

Use ChatGPT as a tool to augment your work, complementing your expertise and decision-making processes.

Professional bio

As a Chartered Accountant with over 25 years of international risk management and corporate governance experience in the private, not-for-profit, and public sectors, Patrick helps individuals and organizations make better decisions to achieve better results as a corporate and personal trainer and coach at Practicalrisktraining.com.

 

Leave a Reply

Your email address will not be published. Required fields are marked *