I used Generative AI to complete day-to-day tasks, here’s how it went

Generative Artificial Intelligence (Gen AI) looks set to change the way we work and the way we play in the decades to come.

The promise of Gen AI is vast, automating tasks that people never thought possible, such as writing emails, image creation, coding programmes, and even generating increasingly lifelike videos.

I’ll say upfront, I’m a writer and a tech enthusiast, so I look at Gen AI with a sense of both awe and scepticism.

It is impossible to understate the impact that Gen AI will have, the capabilities it will bestow, the ways that it has already changed the way we work. However, bear in mind that Gen AI is also not a be-all, end-all solution.

Like all new technologies, we are still working out both the problems with Gen AI and we can make full use of it.

So I spent some time using Gen AI products to complete some day-to-day tasks: writing emails, letters, making grocery lists, traveling home, and improving my hobbies.

It’s been here the whole time

I’m dyslexic and I use a lot of software to help with the condition. Interestingly, much of the software programmes involve Gen AI today.

If you’ve used a voice assistant like Siri or Google Assistant, it's partly trained on Gen AI.

If you wrote an email and the programme gave you a suggestion to finish your sentence, that’s partly Gen AI.

I used to pay US$300 (S$387) for a speech-to-text programme to help me write essays.

That functionality is now available to me, for free, on Microsoft Word, built on the same technology that I paid for.

“But speech-to-text isn’t Gen AI!”

Yes, it is.

Nuance, the company that sold me my speech-to-text programme, was acquired by Microsoft so it could integrate the former’s technology into its offering.

Nuance billed itself as “a leader in conversational AI and ambient intelligence”.

And so you see levels of integration that make many AI processes already part of our everyday lives.

The future isn’t AI. The present is.

Chicken little

Gen AI’s capability has grown exponentially from what we would consider simple, but tedious actions that we all can do, to things that are beyond our native capabilities.

You’ve already seen one example: the cover picture for this article was me generating an AI image for use in this article.

I don’t have the artistic or technical capability to produce such an image on my own.

Let me give you another example.

I like to dabble in sketching, so here’s a sketch of a chicken I drew. I drew it freehand, on my tablet, with a picture for reference.

It’s okay, in that you can identify it as a chicken.

So I went to Microsoft Copilot (its main features are available for free, although with limited prompts), put in the same picture, and asked it to make it more realistic.

So, yeah, that’s better.

And then you can go mad with it, like this Sci-fi bounty hunter chicken.

My chicken sketch took me an hour. The other chickens were all generated in five minutes.

To me, this represents the best version of AI-human combination, the ability to take ideas that individuals cannot otherwise express and give it a voice.

Letter writer

I don’t use AI in my own writing (I do use an AI-based programme to help check grammar though), with one exception.

When I have to write an official letter, to a company or similar, I often use an AI tool (ChatGPT, when I still had a subscription, now Copilot) to help adjust tone and make suggestions.

Happily, I don’t currently have a need to communicate with my bank or my boss, but I played around to generate a few examples of how a letter like that might look.

Think about it, this can be really valuable for those who do not have a strong command of English to communicate their urgent needs or to promote their businesses in today’s digital era.

Another day-to-day use where Gen AI can come handy is to use it to generate a grocery list.

I’m not good at meal planning, or grocery shopping, and it's always a bad idea to head to the supermarket when you’re already hungry.

It came up with a list, priced it, and even had recipes to recommend.

That said, the state of Gen AI at the moment means that I do have to double check the prices and the recipes properly before heading to a supermarket.

On the bright side, I know of friends who use Gen AI to come up with travel itineraries as well, and they find it incredibly useful and time-saving.

Final destination

To experiment how far I can rely on Gen AI, I tried using it to help me find an efficient way to get home.

But I thought, I don’t want to reveal my address, so I generated a route to get to my previous residence, where I lived for about a year.

Essentially, I had stayed there long enough to know the area, but not so long that I’m sure I knew the most efficient way to get there.

Not all Gen AI programmes have access to up-to-date traffic patterns, but I thought it might be able to show a good general pathway.

Let's just say it didn’t go the way I thought it would.

I got odd results that told me to take train lines to stations they did not go to, to take buses from bus stops that those buses did not stop at, and to alight at bus stops that were very far from the ultimate destination.

For example, this is an illustration of the range of final bus stop locations I was given, and the target destination wasn’t even in the centre of the circle.

Prompt and proper

In recent months, I’ve tried to coax Copilot to help me develop a programme or a macro to scrape data to collect statistics for work.

I haven’t got it yet, and I suspect I still don’t know enough to actually get the right prompt.

But Copilot helped me learn enough about the process that I was able to contribute meaningful suggestions about how to improve our statistics collection.

All I’m saying is that Gen AI is not a perfect tool, and users need to be canny in using it, and checking its results.

Because what it can do is really remarkable, if you know how to do it right.

This branded article by Digital for Life is brought to you in part by a gamer who will never be able to afford a top-of-the-line GPU ever again.

Top image via Microsoft Copilot and by writer