The Problems I Want AI To Fix

The Problems I Want AI To Fix

Looking through my RSS feeds, the media releases I receive in my email, and the headlines in the tech press and on social media, there’s one thing that seems to be dominating much of the conversation – artificial intelligence. Some of the applications seem quite interesting, such as the TranslateOne2One. There are others that solve smaller, yet annoying issues, such as the announcement by Slack this morning. But what are the problems we want AI to solve?

Before diving in, I’m going into “Bad Year 10 English Essay Mode” and starting by defining what I think AI is. Simply, it’s a system that can look at what I’m doing, understand it in context and then enhance what I do in future based on that knowledge. For example, if my morning ritual is to wake up, turn the coffee machine on to make my morning heart-starter, and then go to the loo, I want something to know that when I get out of bed to press the switches and turn on the lights for me.

#1 Email

Google is progressing on this front but I want some smarter processing of email. I get about 200 messages a day. Many of these I need to act on quickly, some can wait till tomorrow and others need to be quickly scanned and squirrelled away for that mythical time called “spare”.

I’d like AI to be able to detect which messages need some immediate action and tell me. This could be based on language or the presence of a time or date in the message.

#2 Travel planning

This one seems easy to me but airlines and airports seem to be struggling.

If I catch a plane from Melbourne to Sydney, and that plane is coming from Perth but was delayed on take-off by a couple of hours, what can’t the airlines reliably tell me that there will be a delay?

It happens sometimes but too often, a delay that should’ve been known hours before only reaches passengers when they’re sitting on the tarmac or trapped in a packed terminal. This happened to me a few weeks ago. I was on a flight from LAX to MEL and we were stuck on the tarmac for two hours because a flight from JFK was delayed. That’s a five hour flight – plenty of time to tell passengers and let us sit somewhere more conformable than a cramped economy class seat.

#3 Shopping

One of the things I most hate is when a product I buy is superseded a week or two later, or is discounted just after I make my purchase.

When I shop for a product, I’d like some indication of the product’s refresh cycle and whether there are cheaper prices elsewhere. While there are marketplaces that do this to some degree, I’m sure a smart AI engine could look at years of history and inform us as to when we can make smarter shopping decisions.

#4 Tell me how to have a good day

I’m training for a marathon at the moment (Surf Coast Trail in a couple of weeks so, technically, I’m tapering). I’ve done plenty of long runs over the last few weeks and did the same race last year.

I capture all my training data, wear a fitness tracker, record my sleep data and track what I eat (using MyFitnessPal) closely.

I want all of that, and environmental information such as the weather, to tell me what the optimal training times are for me and how much sleep and what meals will help me get the best possible result from my race.

I also want to know what I need to do to be at my most productive at work.

The data is there – but there’s no easy way to bring it all together and make it actionable.

So, what do you want AI to fix? What challenges do you see out there at home or work that AI could help overcome?


  • I’d like software recommendations. There are thousands of apps, programs and websites that might do what I want for any given task, but most of them don’t work the way I want, and even finding out what they all claim to do is a task in translating marketese to human. I have my preferences, my style of work, my common circumstances. If an AI could take those preferences and the task I’m trying to perform, then give me a list of actually suitable apps, rather than plain keyword searching, I’d be very grateful.

  • I would like AI to smartly organise my electronic files. If I download music, why do I have to go into a downloads folder to move the music to a separate folder. If I download presets for Photoshop, why do I have to manually move them into a specific folder and then install them. If I download an attachment from an email, why can’t it be organised straight into a specific folder based on what the attachment contains? The less time I have to spend searching for things, or organising folders, the more time I have to enjoy them right?

    • This is a good idea. Although, instead of organising into folders, maybe if AI could automatically tag files you download based on its perceived if you need to search for “that document I downloaded from email about work” it can just zone in and identify any file that meets criteria?

  • What you are referring to is automation, not AI. Unfortunately marketing folks have hijacked the term. If you have even a single line of code, it is not true AI. It’s a lost cause at this point. AI fans need to come up with a new term for AI.

  • I think AI needs to focus on boring and tedious problems.

    To me, an odd trend in medicine is the attempted usage of AI for diagnosing illness, especially the “this robot can read a CT and find a tumour better than 98% of Doctors!” clickbait.. one of the bigger problems any health professional faces on a daily basis is the sheer amount of documentation they have to do.

    Instead of trying to solve problems that are more art than science, why not use it to make a problem that has real clinical impact, and would make work easier?

Comments are closed.

Log in to comment on this story!