You’ve heard the hype on how amazing AI is, and then you tried it and were… underwhelmed. Or maybe you’ve read how dangerous AI is, and so you don’t want anyone in your organisation to risk using it.
So, what is the AI reality for people that work in small to medium sized businesses in New Zealand?
What is AI good at, and what are the risks?
What new features have we seen recently that are exciting and useful? And what complete failures have we experienced? Where should you start when considering AI tools that might help your role or business?
- Introduction to AI Webinar: Rhys Kerrigan introduced the AI webinar, explaining the focus on AI tools and their relevance to New Zealand’s small to medium businesses. He mentioned that Ben Fish, the practice lead for digital enablement, would assist with technical questions.
- AI Tools Overview: Rhys Kerrigan discussed the various AI tools available, including large language models and machine learning tools like ChatGPT, Copilot, Claude, and Gemini. He highlighted their applications in summarising documents, improving content clarity, generating imagery, and more.
- AI Tools Introduction: Rhys provided an overview of AI tools, focusing on large language models and machine learning tools such as ChatGPT, Copilot, Claude, and Gemini. He explained that these tools help automate tasks and assist in daily business operations.
- Applications of AI: Rhys highlighted various applications of AI tools, including summarising long documents and emails, improving content clarity and grammar, generating imagery and infographics, and converting outputs to different formats. He mentioned specific tools like Copilot, ChatGPT, and Adobe PDF for summarising documents and emails.
- AI as a Personal Assistant: Rhys Kerrigan outlined the strengths of AI as a personal assistant, such as summarising long documents, improving content clarity, generating starting points for writing, reconciling spreadsheet data, assisting with meeting notes, and converting outputs to other formats.
- Summarising Documents: Rhys explained that AI tools are effective at summarising long documents and emails, helping users quickly understand key points and updates. He mentioned tools like Copilot, ChatGPT, and Adobe PDF that offer summarisation features.
- Improving Content Clarity: Rhys discussed how AI tools can enhance content clarity and grammar, particularly for individuals with English as a second language or those with lower literacy levels. He highlighted the benefits for people with dyslexia and those who use sign language as their first language.
- Generating Starting Points: Rhys noted that AI tools can help overcome the challenge of starting with a blank page by generating initial drafts for emails, documents, and reports. He emphasized that while the generated content may require adjustments, it provides a useful starting point.
- Reconciling Spreadsheet Data: Rhys mentioned that AI tools can assist in reconciling and cleaning spreadsheet data, although users need to understand how to use these tools effectively. He highlighted the importance of providing clear instructions to achieve accurate results.
- Meeting Notes and Actions: Rhys highlighted the integration of AI tools with meeting platforms like Microsoft Teams, which can record, transcribe, and summarise meeting notes and actions. He described this feature as particularly useful for individuals who host frequent meetings.
- AI as an Automation Tool: Rhys Kerrigan provided examples of AI as an automation tool, including streamlining emissions data collection for construction operations, generating automated email text for managed service providers, and building chatbots for retailers to handle FAQs.
- Emissions Data Collection: Rhys described a case where AI was used to streamline the collection of emissions data from suppliers in the construction industry. The automation process saved significant time, reduced reliance on a single person, and ensured consistency in data collection.
- Automated Email Text: Rhys mentioned that managed service providers, including Kinetics, use AI tools to generate and correct time entries and invoice annotations. AI is also used to draft automated email texts for monthly reports, making the communication more personalised and efficient.
- Chatbots for Retailers: Rhys discussed the use of AI to build chatbots for retailers, which handle frequently asked questions (FAQs) on their websites. This automation reduces the workload on customer service teams and improves the accessibility of information for customers.
- Weaknesses of AI Tools: Rhys Kerrigan discussed the weaknesses of AI tools, such as expecting too much, poor branding, automating broken processes, skipping human review, writing in a unique tone of voice, lacking human empathy, and being unclear when working on data.
- Expecting Too Much: Rhys cautioned against expecting AI tools to solve all problems instantly. He compared AI to a clever intern who needs context and guidance to perform tasks effectively. Users should manage their expectations and understand the limitations of AI.
- Poor Branding: Rhys noted that AI tools, particularly Copilot, struggle with applying branding and templates correctly. While improvements are being made, users should be aware that AI may not always adhere to specific branding guidelines without additional setup.
- Automating Broken Processes: Rhys emphasized that AI cannot fix broken or unclear processes. Organisations need to address process issues first before implementing AI to ensure effective automation. AI should be used to enhance well-defined processes, not to fix flawed ones.
- Skipping Human Review: Rhys highlighted the importance of human review and oversight when using AI tools. He stressed that AI-generated content should be checked for accuracy and relevance, as AI can make mistakes or produce incorrect information.
- Unique Tone of Voice: Rhys mentioned that AI tools may not capture an individual’s unique tone of voice, which can be frustrating for users. While AI can generate content, it may require adjustments to align with the user’s writing style.
- Risks of Using AI Tools: Rhys Kerrigan highlighted the risks associated with AI tools, including privacy concerns, accuracy issues, copyright considerations, bias, reliance on AI, and hallucinations where AI makes up information.
- Privacy Concerns: Rhys warned that free public AI tools like ChatGPT use the data provided to train their models, potentially exposing sensitive information. He recommended using tools with clear governance and security measures, such as Microsoft Copilot, to protect data.
- Accuracy Issues: Rhys explained that AI can produce convincing but inaccurate information. Users should verify the accuracy of AI-generated content and understand the sources of the information to ensure reliability.
- Copyright Considerations: Rhys advised caution when generating imagery with AI tools, as the content may resemble copyrighted material. Users should be aware of copyright implications and ensure that generated content does not infringe on existing copyrights.
- Bias in AI: Rhys discussed the potential for bias in AI tools, which may be influenced by the data they are trained on. Organisations should implement policies to monitor and address bias in AI-generated content to ensure fairness and diversity.
- AI Hallucinations: Rhys described instances where AI tools make up information, known as hallucinations. He provided an example of AI generating additional columns in a data table that did not exist. Users should be vigilant and prompt AI tools to use actual data to avoid such issues.
- Starting with AI as a Personal Assistant: Rhys Kerrigan provided tips for starting with AI as a personal assistant, such as learning prompting as a skill, collaborating with the team, saving best prompts, starting small, setting guardrails, and protecting data.
- Learning Prompting: Rhys emphasized that prompting is a new skill essential for effectively using AI tools. He advised users to practice and refine their prompting techniques to obtain accurate and useful results from AI.
- Collaborating with Team: Rhys encouraged teams to collaborate and share knowledge about AI tools.
- Starting with AI as an Automation Tool: Rhys Kerrigan advised on starting with AI as an automation tool, emphasising the importance of understanding the problem, defining the issue, clarifying outcomes, selecting the right tools, and seeking help if needed.
- Closing Thoughts on AI: Rhys Kerrigan concluded the webinar by stating that AI is here to stay and emphasised the importance of upskilling, fixing processes before implementing technology, building confidence through training, and seeking help when needed.
- Questions and Answers: Rhys Kerrigan and Ben Fish addressed questions from participants, discussing the importance of due diligence when selecting AI tools, concerns about privacy and data protection, and the potential risks of using certain AI applications.