Trust, Hallucinations and Checking AI Output

“By far, the greatest danger of AI is that people conclude too early that they understand it” —Eliezer Yudkowsky, AI researcher

AI is everywhere today, and there are many exciting claims about what it can do to help us be more productive. But, is this just hype, or are there aspects of AI that can improve our productivity? 

That’s the question I am answering today. 

Links:

Email Me | Twitter | Facebook | Website | Linkedin

The Hybrid Productivity Course 

Get Your Copy Of Your Time, Your Way: Time Well Managed, Life Well Lived

The Working With… Weekly Newsletter

Carl Pullein Learning Centre

Carl’s YouTube Channel

Carl Pullein Coaching Programmes

Subscribe to my Substack 

The Working With… Podcast Previous episodes page

Script | 407

Hello, and welcome to episode 407 of the Your Time, Your Way Podcast. A podcast to answer all your questions about productivity, time management, self-development, and goal planning. My name is Carl Pullein, and I am your host of this show. 

You may have noticed AI is everywhere. Our favourite apps seem to be adding more and more AI capability with each new update. And then there’s almost every video and article on productivity warning us that if we don’t get on board with this, we’ll be left behind on the scrap heap. 

It’s also an exciting time, and there’s no doubt that things are changing, and people are finding new ways to use AI to help us do our work. 

But beyond the hype, how are current AI models really helping with productivity, and what will this mean for us as we try to manage our time in the future? 

That’s what I am looking at this week, and to get us started, let me hand you over to the Mystery Podcast Voice for this week’s question. 

This week’s question comes from Chris. Chris asks, Hi Carl, I haven’t heard you talk much about AI in your videos or articles. How do you see AI helping us with our time management and productivity in the future? 

Hi Chris, thank you for your question. 

The reason I have not written or spoken much about AI is that I am waiting to see where it settles down. 

Currently, it’s hard to work out what is true and what is pure hype. I saw a lot of noise about OpenClaw—an AI-type personal assistant that, if you give it access to your computer, can do a lot of things, such as make appointments for you, book flights, sort and reply to your emails and much more. 

That was certainly interesting, but once I discovered that I would need to hand over all my passwords and credit card numbers to OpenClaw, I lost interest. 

Call me old-fashioned, but I’m not comfortable giving up my passwords, credit card and banking details to a third party. Certainly not one that could be hacked very easily. 

Last year, I read Dominic Sandbrook’s series of books on British history from 1956 to 1982. That period covered some very interesting developments in technology, from the dawn of the nuclear power age to the introduction of the personal computer.

In the late 1950s, it was predicted that we would all be driving around in nuclear-powered cars and that our homes would have their own nuclear power generators that would only need recharging every 10 to 20 years by the end of the century.

Hmm how did that work out? 

To better answer your question, Chris, I stepped back and looked at how I am using AI today.

My main use of AI is searching for specific information. In a way, AI has replaced how I search the internet. I use Google’s Gemini, and it is fantastic at collecting the information I want. 

No longer do I have to open multiple websites to try to find the information. This has significantly reduced the time I spend going down rabbit holes looking for something specific and being pulled down holes I never intended to go. 

I also use AI to generate subtitles and timestamps for my YouTube videos. Without AI, these jobs would take hours. AI can do it in minutes. 

I use Grammarly to spell-check my writing, and I believe it uses AI in the background to suggest how sentences are written. 

I rarely accept Grammarly’s sentence suggestions. It seems to destroy my voice and turn sentences into bland perfections that lack resonance or feeling. 

Beyond that, I am not knowingly using AI for anything else.

I asked my wife how she is using it. My wife’s a full-time student, studying physical therapy, so she’s learning a lot about human anatomy and medical terms. 

She’s using AI to simplify complex concepts. She also occasionally uses Google’s Nano Banana to generate graphics for her presentations. 

So, if I look at how AI might help us with time management and productivity in the future, it does look like there will be some aspects of our work that AI can significantly speed up. In my case, generating subtitles and time stamps for videos is a great example. 

However, when it comes to managing our calendars and task lists, I’m not sure you would want AI getting involved. 

One thing I’ve always been acutely aware of is that much of what makes us feel overwhelmed is the sense that we have no control over how we spend our time. 

We have calendars full of meetings, and sometimes we find ourselves double and even triple-booked. And then we have long lists of to-dos in our task managers with no sense of when or even how we will ever get that work done. 

At best, AI may be able to break down those tasks into what it thinks are manageable chunks, but that won’t take into consideration how you are feeling physically, whether you slept well last night or had a rather heavy lunch with an important customer. 

AI can certainly suggest ways to manage your tasks and calendar, but you will still need to show up to those meetings and do that work. 

Yet that will inevitably leave you feeling less in control of your time. Particularly if you use one of those AI-enabled calendars that suggest when you should be doing something. 

What happens if you disagree with the suggestion, or you cannot make it? You feel guilty, or you start to think something is wrong with you. 

Yet, there’s nothing wrong with you. You’re human, and you are going to feel tired sometimes or not in the mood to do that type of work. 

The one area I would say you want to avoid AI getting involved in is how you manage your time. That should always be your responsibility and choice. 

The idea that a computer tells you what to do and where to be is scary. Deciding what you do right now is what makes you human. You’ve chosen to listen to this podcast at this time. AI would likely tell you that, rather than listening to this podcast, you should be finishing that report you’ve been trying to finish all week.

I also read about the excitement over the idea that AI could reply to your emails for you. Hmm, for me, that is a red line I will not cross. 

Call me old-fashioned, but I believe that if someone has taken the time to write to me, I have an obligation to reply personally. That is just basic integrity.

Now, it is true I don’t reply to all emails. I don’t respond to spam emails; for example, I simply delete them if they get through. How hard is that? 

I’m fortunate that I’m old enough to remember several technological advancements. It started with the Internet, then email, the smartphone and cloud computing. 

I cannot remember a technology being forced upon us, but it feels like AI is being forced on us, whether we like it or not. 

And then there are the frightening ads that claim if you are not on board with using AI, you will be left on the career scrap heap by the end of the year. Nobody needed to do that with smartphones or email. 

Companies, focused on making the technology user-friendly in such a way that we all wanted to adopt it eventually. The fear-mongering I see around AI makes me deeply suspicious of it. Why do they need to do that? 

Perhaps that question is for people better qualified than I am. 

Anyway, AI is here, and it’s not going to go away. 

Where I think AI will be a huge help to us is in repetitive, mundane work. I mentioned that I use AI to create subtitles and timestamps for my YouTube videos. That’s been a huge time saver for me. 

But if you follow my email processing system, you will find that you are faster than AI. I can clear 80 emails in my inbox in less than 10 minutes. It’s also important that I do this, as I want to get a heads-up on my day. To know if there are any emergencies, what I want to read later and what I can delete. 

What AI would do is categorise your emails between what it thinks is important and what is not. Trust me, you will do a far better Job of that than AI will. 

The problem here is that you will not trust AI 100%, so you will still go through the emails it thinks are not important, just to check that it got it right. 

And that’s a big problem with AI today, although I accept that in time this may change; people don’t trust it, which is a good thing, as AI can hallucinate and give you incorrect information. This means you spend time coming up with the right prompt, get the answer, and then have to check that it’s correct. 

The question then is: did it really save you time? 

I am monitoring AI carefully. I know that in time, it will bring us some productivity benefits, new technologies always do. But there are a few areas where I won’t use AI personally. 

Writing emails and answering user comments. That’s a personal integrity thing to me. Your principles should tell you that. 

Managing my calendar. That’s another personal thing, and giving control to any outside influence would always be problematic at a human level. 

Creating content. If you’ve read an AI-generated blog post or watched an AI-created YouTube video, you can tell. Large Language Models will always default to the average, not just in the content, but in the words used. It’s horrible, and nothing unique will ever come from it. 

And finally, deciding what I will do at a task level and when. That’s another one that, as a human, I will retain control. I had scheduled to write this podcast script at 11:30 today, but I had a cancellation at 8:00 am, so I switched things around. 

I could have gone back to bed, but I felt great, so I decided to get on with this podcast script. My choice, made in the moment. 

Thank you, Chris, for your question and thank you to you, too, for listening. It just remains for me now to wish you all a very, very productive week.