My Account  

    • Be careful – AI can be a time waster!

    My poor experience of AI

    I have just finished writing the 47-page Change Capability Report describing the insights from the global, annual Change Capability Survey. Stupidly, this year I thought this mammoth task would be made easier by using AI – I was very wrong!

    My experience of generative AI platforms like claude.ai and chatgpt.ai was that they made things worse. I expected these tools to make the review of the data easier, because of the speed of analysis and the volume of information that I can share and ask questions about. There is no question that AI has amazing processing power but…it is the results that are the problem, not the ability to analyse that is wasting my time.

    For every question I asked, I would get pages of notes in reply. Instead of saving time, I wasted time trying to read everything these platforms produced for me. I gave clear instructions about short, punchy paragraphs, I requested it to report one trend at a time, and it would still come back with 8 pages of text.

    Even when I sent further prompts demanding shorter, punchier text, it would only partially comply and cut the text to four pages instead!

    AI wasted my time

    The scale of the replies had these impacts:

    1. I needed a lot of time to read everything so I wasn’t gaining any time saving efficiencies.
    2. The answers were so unspecific, I started to lose my own train of thought.
    3. The answers had lots of repetition, and were not usable unless they had a lot of editing, which took even more time than writing my own thoughts from the start
    4. The text would include large amounts of “extra” text that padded out the answers, but did not add any value.
    5. This extra text came from all over the internet, some of it very out of date.
    6. Sometimes the answers were just plain wrong – I know in AI we are supposed to call this hallucinations but it doesn’t make it any better, the answers were unusable because they were factually incorrect.

    I tried reading the replies aloud to see if this would keep me more focused, but instead I could hear how boring I sounded as I droned on, page after page. It reasoned that if this is how I felt, with all my passion for the subject, I can imagine that anyone reading my work would feel similarly bored and uninspired!

    AI doesn’t have empathy, it cannot read human emotions and really understand our perspective. Perhaps that is why it continued giving me long, rambling answers often containing inaccuracies to a time-poor expert in her field.

    Criteria for AI that makes my working life better

    If AI is to make things easier, I need it to perform like this:

    Respect me

    Let me tell it what my situation is and have that respected – not ignore what I have said and provide generic answers from a general scraping of all the content on the internet. When I delivered a webinar on how to use AI for change recently, the audience and I talked about what AI should be called, and one of the answers was Assimilated Intelligence – and that is a pretty good description of what it does. It assimilates or approximates an answer from multiple sources, but the problem is many of those sources are not accurate, not respected and dilute rather than support my argument.

    Respect my instructions

    I want AI that will “read the room” and understand my needs. If I tell it how I want an answer formatted, I want it to respect my request. For example, if I ask for the short-form first, it should deliver 3-4 sentences, no more. If I want more detail I will ask it.

    Use quality sources

    My use of AI cannot be thoughtless. If I want meaningful answers I must specify some of the sources I want it to use, and explain why I believe these are quality sources, so it can look for others with the same qualities. For example, I am very cautious about any leadership guidance that was written before Covid because since then, we have had the seismic leadership shocks of hybrid working and an AI enabled world. I am not saying everything written before 2020 is not useful, but I am cautious about the context in which these materials were created.

    Finding AI that meets these quality criteria

    One of the reasons I am so proud of ChangeabilityPro® is that it delivers the right answers at the right time. We do use AI based search tools, so you can search for answers that match your situation, but those answers are specially curated to ensure they are accurate, relevant and easy to apply. We do not do a general scraping of the internet, we have contained our materials to our own library of change specific resources, updated every 2 weeks.

    ChangeabilityPro® is filled with the latest techniques to address how we are managing change today. It does not churn out long form explanations from theorists who are long dead and whose work was never designed for the volume of inter-dependent transformations taking place in organisations right now.

    Each answer comes fully formed, with a video and instructions for how to apply the technique to your work, including checklists of actions to take, and information for you to use, guidance notes giving you a step-by-step guide for what to do and templates for you to tailor to your own change.

    Conclusion

    AI has its uses, but you will not get the best out of it unless you use it thoughtfully and ensure that it meets your needs. Establish your criteria for success and use these to get the best out of generative AI tools.

    Use the answers that you get as input to your own ideas, not as replacements for your thoughts. Only you fully understand your context and your needs. AI is not human, it doesn’t understand the nuances of how we interact with each other and what other humans need.

    Ultimately, trust your own judgement and use AI as one of many sources of your own intelligence.