Can AI make my life easier? I spent a week living and working with chatbots to find out

2025-04-26T05:00:00Z

Artificial intelligence isn’t going to take over our lives just yet, but it can make certain jobs quick and easy

Artificial intelligence wrote this article. That is a lie actually; I thought AI might write this article, but after several attempts with different models, the end result was peppy, relentlessly positive and … not good.

If you want something done properly, they say, do it yourself. For all our fears about AI taking our jobs, there are still some things that it – fortunately, or unfortunately depending on the task – can’t quite do yet.

Three years ago, most of us had little idea what ChatGPT was, or why we should care about AI. Fast forward to 2025 and it seems everyone wants in on the AI game. There are chatbots managing customer service, a kind of high-tech automated phone system steering you to the right person. There are virtual assistants offering help on every app you open. We can’t get away from AI that apparently exists just to make our lives easier.

But does it?

We are naturally suspicious of the technology. AI gone mad has been the basic premise of numerous dystopian futuristic nightmarish movies. In 2001: A Space Odyssey, Hal took over the ship and went rogue. Terminator not only started with a sentient AI taking over the world’s defence systems, it built an army of robots to finish the job. And the Wachowskis – most famous for the Matrix series – managed to eke out far more movies than necessary out of the whole “robots at war with humans” theme. If we are dependent on Hollywood for our views on AI, it is no wonder we are all terrified of ceding control to the machines.

Still, there are plenty of good ways to use AI, and so I am tasked by The Irish Times to document how it can be used in day-to-day life over the course of a normal week.

A bit of thought reveals that I am already using it regularly, even if I didn’t realise it. Take my email for example. It already suggests replies, organises itself into categories, and bumps priority emails back to the top of my inbox. As a journalist I use AI to transcribe interviews, saving several tedious hours every week. And I probably interact with chatbots regularly on websites without realising.

Then there is my mobile phone – or phones, if I’m being honest. Both my iPhone and Android have AI built in now. Android’s AI has useful features such as Circle to Search, which allows you to tap or circle anything on the phone’s screen to search for it online, or you can have a full-on conversation with its digital assistant Gemini.

I ask Gemini how to deal with my 10-year-old’s newfound ability to roll her eyes at me. Then I make her listen to the answer

Apple Intelligence is just getting started. For phone users, it is only available on the iPhone 15 Pro and the new iPhone 16 series; luckily I have the latter. It can give me AI-powered writing tools to proofread or change the tone of an email or document, or help me create my own custom emojis. You can also use the camera to search for images online. I’ve been using a preview version of the software for a few weeks, but have rarely dipped into the AI features so far.

This week will be different though. I’m going to give AI a proper shot, using it for every day-to-day task I can think of.

I decide to start small. An email reply here, a proofread document there. But it quickly snowballs as I find more uses for it. What about some holiday recommendations? Can it make me a better writer? How about life advice? Before I know where I am, AI has infiltrated every part of my life.

Monday

Three years ago, most of us had little idea what ChatGPT was, or why we should care about AI. Illustration: iStock

On Monday morning, I check my personal email on my iPhone to find Apple Intelligence has helpfully arranged everything into conversations, with short AI-generated summaries to flag the most important bits. It has also added a priority section at the top of my inbox, just in case I missed that very important deadline for a child-related activity coming up. The day’s emails are dealt with in a few seconds; replies are proofread by AI. One useful feature is the automatic follow-up prompt – if I’ve asked a question that hasn’t been answered by return email, the system prompts me to chase it up. It is like having a personal assistant on hand.

I have a feeling I rarely experience: almost organised.

When I open my work inbox, that warm and fuzzy feeling is instantly squashed. There are a few hundred unread emails, and in this case, I have no AI to help me. I am all my own.

It doesn’t have to be this way. Microsoft – which powers Outlook – was a relatively early adopter of AI. The company has ploughed billions into OpenAI, the maker of ChatGPT, and integrated the software into many of its popular products. There are CoPilots – or AI assistants – for almost everything. They can check your email and prioritise the most important messages, compose documents for you, summarise video content, and – best of all – sit in on Teams calls, take notes on your behalf and send you a summary when everything is done, picking out tasks and queries that are specifically addressed to you. Provided you – or your employer – can pay for it, that is. CoPilots cost actual money, and more importantly in my case, require permission from the powers that be to install them on your work system.

[ Irish authors seek to stop Big Tech ‘scraping’ their work for AI amid copyright breach claimsOpens in new window ]

For now, I’ll have to read my own email. How boring.

But wait. A strange email claiming to be from a password manager that I used in the past pops into my inbox. Apparently my login was leaked and I need to change my details. I’m naturally suspicious of anything like this, so I ask Siri – Apple’s voice-activated AI assistant – if this content is likely to be a fake. Siri can’t answer, offering instead to send the screenshot to ChatGPT. I agree, and a screenshot is fired off to OpenAI’s assistant. It correctly flags it as a potential scam – I’ve already checked the sender and it doesn’t match up – but it also raises a few points that I didn’t consider.

Just in case, I check a few more emails, and it seems that ChatGPT isn’t just throwing out a few generic security tips. Each time I ask, it highlights different elements of the message that indicate whether it is fake or genuine. Score one for AI.

I have a few articles to write, but The Irish Times isn’t one of the newsrooms that has adopted AI to write basic stories. I don’t think now is the time to stage a one-woman AI-fuelled charge, but I ask ChatGPT to write a persuasive argument to convince the newsdesk that it is a good idea. I file that one away for later.

I already use some AI tools occasionally at work. I have tried using AI for a first pass at transcribing lengthy speeches at press conferences or public events, which can save a lot of time. But I still need to check the text for accuracy myself. Also, if the transcription service uses cloud technology, you have to trust that it will keep tight control over your data. The best way to make sure everything stays confidential is not to upload it to a cloud platform to start with, so I wouldn’t use it for sensitive interviews.

Some tools – Otter AI, for example – can also attend virtual meetings for you and take notes, but the silent assistant in the (virtual) corner isn’t for everyone.

Speaking of time saving, what about the domestic side of life? I waste at least an hour each week reminding Child Number One that she needs to empty the dishwasher, that it was her own suggestion she take on the job, and her extra pocket money hinges on it.

[ No escape for professions from rolling revolution that is artificial intelligenceOpens in new window ]

This week I’ll leave it up to AI to decide who does what around the house. Gemini suggests a family meeting and making things “fun” with a chore chart. I can think of better ways to have fun, but we give it a go.

CoPilot creates a chart, but I can’t follow what it is trying to tell me; it is a mess of checkmarks. ChatGPT vindicates our current approach by suggesting the 10-year-old helps with loading and unloading the dishwasher. It also creates a full list of daily, weekly and monthly chores, and assigns them.

As a follow-up question, I ask Gemini how to deal with my 10-year-old’s newfound ability to roll her eyes at me. Then I make her listen to the answer.

Tuesday

Tuesdays are early starts here. Before going to bed I asked various AIs – Google Gemini, Apple Intelligence, Claude and ChatGPT – what was the best way to get a good night’s sleep and wake up refreshed at 6am.

Apparently I’m doing it all wrong. My screentime is out of whack with all their recommendations, I exercise at the wrong time of day, and I am a chronic snooze-button abuser. I never take a daytime nap though, and I don’t have a TV in my bedroom, so Gemini gives me some points for that.

I use Otter AI to transcribe a product presentation. It does a decent job, and fast. To eliminate the cloud aspect, I also try a Pixel phone with Google Recorder, which processes everything on the phone. It also transcribes it reasonably accurately.

In the afternoon, the robot vacuum is let loose on the mess around the house while I am out. It used to get stuck on cables abandoned on the floor, or eat a sock or two on the way around. Now it has AI “obstacle avoidance”, so it makes its way around anything in its path, and helpfully marks a symbol on a map, which I can check on my phone. It could be a shoe, others it’s a power cord. The one that makes me stop and pause is the poop symbol. I’m almost afraid to go home.

Wednesday

‘It turns out AI is not that funny’. Illustration: iStock

As part of the “new me” thing I kick off every couple of months (and subsequently abandon), I’ve decided I need a new fitness plan. Forget about the gym membership I pay actual, real-life money for; I feel now would be a good time to put AI to the test. I ask several apps to create a running plan to help me run a marathon. Gemini comes back with a 16-week programme that involves four runs a week, increasing the time spent running week by week, interspersed with rest days and cross training. ChatGPT opts for 20 weeks, and goes for a distance-based training plan. Claude gives me a pre-programme checklist to go through, including getting medical clearance before starting, and a mixture of distance and time-based training.

I’ll never actually do a marathon mind you – it is just one of those bucket-list things. and I suffer from FOMO. (Gemini, how do you prevent FOMO?)

Health isn’t just physical, of course, and people are using AI for everything from companionship to therapy. I try out a few conversation apps, with mixed results.

I have a chat with an AI-generated Mario. I tell him Princess Peach could probably save herself if he just left her to it. He doesn’t take that well, insisting he is the only one who could save her.

Since conversation isn’t going great, perhaps AI can help me out with some jokes. There is an ongoing battle in our house to find the worst dad-joke possible, but the one rule is that it still has to make people laugh. Claude offers up a few, the best of which is: “Did you hear about the guy who invented the knock knock joke? He won the no-bell prize.”

It turns out AI is not that funny.

But it redeems itself later in the day. Small children are relentless question machines, and AI exists to answer questions. If I can’t answer the left-field queries that my seven-year-old lobs at me, usually while I’m driving, he immediately addresses his question to Gemini. “How do you say underpants in Irish?” Gemini answers correctly – fobhríste – but absolutely mangles the pronunciation.

If my parents had AI back when I was a (very inquisitive) child, their lives would have been much easier. My Irish pronunciation might have been a lot worse though.

Thursday

The week is starting to get to me. I have a twinge in my back from sitting too long at my computer. “Gemini, give me some exercises to help relieve back pain.” It thinks for a moment. Because I am using the voice interface, it doesn’t direct me to websites showing the exercises, preferring instead to give a brief audio description. That does not work for me. At all. I need visuals, and preferably a short video showing me what way I put this arm and where this leg goes. I am utterly confused, and still in pain.

I decide I want, or rather I need, to organise my life. This feels like the perfect task for AI. But Gemini clearly mishears me, and launches into a soothing “you’re not on your own” spiel. It’s comforting, but it doesn’t solve my original problem.

Apple Intelligence fares just as badly. A plaintive “organise my life” gets zero response from Apple’s digital assistant. When I repeat it, things take a turn for the worse.

[ Teachers seek indemnity from legal actions over students’ improper AI use in Leaving CertOpens in new window ]

“If you think it could be serious, ask me to call emergency services or someone you trust,” it offers. I mutter something at it, expressing my displeasure.

“Do you want me to use ChatGPT to answer that?” I don’t think it’s being sarcastic, but it is hard to tell. Then it disappears.

At this point, I’m ready to throw in the towel. Which AI can compose the best email to get me out of this assignment, without putting me on the Magazine editor’s hit list?

I decide against sending the email. Instead, I ask Apple Intelligence to suggest a good relaxation routine to deal with work-related stress. In true executive fashion, it immediately pawns off the request to ChatGPT (but it asks first).

It suggests, among other things, a gratitude journal. Because nothing says “relaxation” more than (*checks notes*) more work.

Instead, I use Apple Intelligence’s “genmoji” feature to create a personalised emoji summing up how I feel right now – my head, explosions, and fireworks. That amuses me, so I do a few more, plus some stylised versions of family members with a new app called Playground. No one would mistake these for realistic images, and there are limits to what it will create – you can’t create obviously offensive images, for example.

Friday

For all our fears about AI taking our jobs, there are still some things that it – fortunately, or unfortunately depending on the task – can’t quite do yet. Illustration: Getty Images

After yesterday’s experiments with AI images, I decide to explore further. Which AI will let me create photorealistic images with few guardrails?

Clearly not Apple Intelligence; it sticks to more artificial looking images by choice. But others happily take up the challenge, within reason.

Open AI’s image-creation tool Dall-E will create some images – Jeff Bezos riding a Blue Origin rocket and Elon Musk in a broken down Cybertruck, for example – but won’t create anything more offensive.

[ When AI images become ‘an insult to life itself’Opens in new window ]

I ask Grok – Elon Musk’s xAI creation – to make an image of Musk; it creates a photorealistic image of the billionaire. Then I ask it to make him look like a clown, which it also does. This is amusing me, but it feels like a bad use of resources.

Rory McIlroy is playing this week at the Masters. ChatGPT puts his chances of winning at 15-20 per cent, noting his struggles in previous outings at Augusta, the mental toll of the career Grand Slam he has been chasing, and the fact that he has been so close on a few occasions – 2022 as a runner-up and his collapse in 2011.

“Historically, the course has been tricky for players who rely too heavily on their driving distance without precision.” I feel like there’s a rebuke for Rory in there, from an AI that has handled fewer golf clubs than I have. I’m mildly offended on his behalf.

“I’d say that Rory is always a contender, but winning depends on several variables falling into place. If he’s in good form and his putting and short game are on point, he’s definitely a threat. However, Augusta’s challenge – with its unique conditions and mental pressure – makes it harder to predict.”

I feel a little guilty about using AI to do this; there is the carbon footprint of a frivolous AI request to take into account, but also I’m finding it hard to work up any enthusiasm for watching the Masters, even if Rory McIlroy is in with a shot of winning, because I find golf so dull I’d rather watch paint dry. According to ChatGPT, that can take up to two hours, and the Masters goes on for days.

As we now know though, AI got it wrong, because Rory did indeed win the Masters, which even my cold, golf-hating heart can appreciate for the five minutes of it I watched. (The final few shots of the play-off, to be exact.)

Saturday

Saturday is usually absolute chaos in our household. There are swimming lessons, football matches and birthday parties to fit in, and an upcoming one that needs to be planned in our own family.

This is where the big guns need to come out: Notion AI. If you want to get organised, Notion AI is the way to do it. It has templates for everything, from work and home life to studying. Plenty of them are free, but others charge, and the prices seem a bit odd. Would you pay $5 to download a tracker for this year’s Formula 1? How about $10 for a “concert memories hub”? You might feel more inclined to invest $3 in a flashcard template for learning new languages, or $5 to track your monthly expenses. I stick to the freebies: a task planner, an advanced to-do list and a diary. Plus a journal to write down my experience with AI as I go.

Come back in two weeks and ask me if I’m still using them all.

I’m helping a family member with house hunting. Notion has a home-moving template for when he is ready to go, but first he needs to buy a property in a market that resembles the Hunger Games. Perhaps AI could suggest a winning bidding strategy? Claude tells me I’ve reached my limit for messages. CoPilot has some good suggestions, which chime with Chinese-created DeepSeek’s recommendations (all of which I’ll keep to myself of course, lest I give up my competitive edge).

Sunday

Sunday starts early, with Apple Intelligence giving me directions (via Waze – a live map from Google) to the beach and sauna for an early morning family swim. Not that I need directions, but it is handy to know about unexpected traffic on the way.

Over coffee afterwards, the subject of AI comes up, specifically Meta AI, the new AI system on Instagram, Facebook, Messenger and WhatsApp. Irish users might have noticed the blue circle appear on your WhatsApp recently, meaning you can now add an AI assistant to your group chats, if you feel the need. The downside: there is no way to disable it.

[ What is Meta AI and why is it in my WhatsApp?Opens in new window ]

I suggest to my dad that he could use it to help plan his upcoming holiday. He threatens to delete his WhatsApp account rather than be forced into using Meta AI. A few days later, Meta AI disappears from WhatsApp, but it is a temporary reprieve – a technical glitch, and within a couple of days it is back.

Apple’s email summaries remind me that I need to pay for my own upcoming family holiday to Spain. That is accurately tagged as a priority mail too, which is helpful.

I need a few recommendations for local restaurants, family-friendly activities, and other amenities within walking distance of our accommodation. For that I turn to GuideGeek, an AI-powered chatbot that works on Facebook Messenger, WhatsApp and Instagram. I ask it to give recommendations of things to do in Alcudia for a family of eight, with five adults and three children, one under five. Lots of steps are out, for example, which GuideGeek takes into account when it recommends a stroll around the old town and a trip to a local park. It also gives some useful information on using cash or card, and the practices around tipping.

Could I have got all this from a search engine? Probably, but it would have taken longer to compile the information, and it wouldn’t have been offered up in a very easy-to-access conversation thread on my smartphone.

That reminds me, I need to practice my Spanish. I’ve been using Duolingo for more than a year, and while I can read the language, I still go into panic mode if someone speaks Spanish in real life to me. The flashcards in Notion won’t do it. I need to practise speaking, so I turn to AI for a quick conversation. Both ChatGPT and Gemini have voice interfaces, ideal for practising my limited skills without making a show of myself.

The conversation is less than flowing, once we get past the customary pleasantries. Then somehow I manage to switch the voice input on ChatGPT to Welsh (I think). Necesito ayuda.

Verdict

Over the week, the tools I was most likely to use were the ones built into the devices and systems I was already using. For example, while Apple Intelligence is still getting started here, the writing tools for proofing and changing the tone of emails were very useful – especially when I needed to send a similar email to different people, with a tweak from a polite personal request to a more professional-sounding communication. You simply tap and hold the screen and select writing tools, choose your preferred style of communication, and the AI does the rest. Gemini became my go-to when I needed to ask a quick question, such as the time I remembered only three words from a Yeats poem – Google couldn’t help, but Gemini delivered.

Some tools are helpful, while others seem overly complicated. And plenty of the tasks I set for it over the week could have been done with non-GenAI search engines and a bit of thought.

So, is AI going to take over our lives? Probably not – yet. And we are a long way off sentient technology taking over completely.

But in the meantime, I’ll still say please and thank you. Just in case.

Best everyday AI

Google Gemini. It is built into the phone, so it is always where you need it and easy to access. Gemini Live, the conversational interface, lets you speak to Gemini like a real person – interrupting, adding new information, changing your mind mid-sentence.

Best AI for research

NotebookLM, Google’s research assistant, gets top marks here. You can upload up to 50 sources of information to each notebook, and Google’s AI will create study guides, pull out a list of frequently asked questions, and summarise the main points. It will also create an AI-generated podcast, where two synthetic hosts will discuss the topic in detail. And it is free – for now.

Best for travel tips

GuideGeek has an AI bot that will help you plan out your holiday, making suggestions for your chosen destination and providing you with a decent itinerary. The chatbot works on Messenger, WhatsApp and Instagram.

Best for organising

Notion AI has plenty of templates to organise your work, home and education life. You can spend a fortune on templates, or you can stick to the freebies – or create your own.

Read full article

Auto-posted from news source