Dearest readers, before I continue, yes- I know what you're thinking right now. Miss Z is trying to wear so many different hats, even in the areas she couldn't possibly be familiar enough with to discuss them beyond the level of casual conversation. At first glance, how could I possibly raise some valuable points about something highly technical like AI? Well, I'm still going to do it anyways, and you tell me how I stack up. Initially, I wasn't interested in it. Then, the media and techies and everyone else started talking about it so much. I still wasn't interested. Then, I started hearing claims that were so dough-eyed, overblown and completely unrealistic as to what AI is supposed to do for us (and to us) in such a short amount of time from now, that I decided to finally look into it myself and see what all the fuss is about. While I struggled to understand the core of how AI is created and trained, I understood very well how it has been applied so far in different industries. Having gone through a lot of information and looking at it as an entrepreneur (yep, best way of putting it haha) and a consumer who expects things to truly work and deliver results, I am really not amused. Let me share with you why, and what questions still run through my mind. It may be too basic for some of you techies out there, whereas for others it may actually may you reevaluate the next AI puff piece you read or watch.
Let's start with cars. To quote Morty, jeez oh man! (only saw a few episodes so far, still working on understanding all the references haha) We have media blaring out how truck drivers will lose their jobs, taxi and delivery drivers too, how uBer is teasing driverless flying taxis, and my oh my it just doesn't stop. I think these were written by people who don't drive (especially in Toronto, the 6th worst place for commuters- IN THE WORLD, and I don't think it's simply because we don't have AI drivers yet). I'm not much of a driver, to be honest, but let's take a quick look back at the first Tesla fatality. The car didn't understand that a speedway with an intersection had a big truck going through it and the car smashed into it at full speed. No healthy, well-adjusted and focused human driver would make that mistake! Sure, I know Tesla says you need to keep the hands on the steering wheel and the car drives itself only under certain conditions. But, can we take a step back and really look at what's being done to us, the public? We are told these cars have an AI self-driving mode for which you need to keep your hands on the steering wheel and it only works under certain conditions but even then not well enough. Would you accept this level of compromise from your current car? How about your current smartphone? No, you wouldn't- and neither would I. Oh, but it's assisting us with our driving. Hmm. If I had an assistant like that who was so limited and prone to fatal errors, I would be bogged down by stress so much that I would be much better off doing the work myself. Besides, why let my own skills for performing certain tasks go bad because I outsourced those tasks to a crappy assistant? Oh, and when we look at alternatives, there isn't much to see. There's uBer, Waymo, Alphabet and other companies competing to make true self-driving cars a reality. Well, the AI can have fun dealing with Toronto flash rains and flooded roads, Nunavut winters and snowed in roads, black ice situations, low visibility, and human errors of other drivers who are still driving their own cars. Finally, do people who work on creating and teaching self-driving AI's actually drive on a daily basis?! If techie stereotype are to be believed, they most likely don't, or don't drive well at all. Simply put, I don't even think that conventional approaches to programming software are going to cut it here. We need something new and incomprehensible in the mix to make this one a reality. Truck drivers of the world can sleep soundly tonight.
Data analysis, whether for marketing, finance or other fields, has been the buzz term everywhere for a long time now. Every business that wants to stay competitive is now hungry for data, as analyzing it allows them to do more things right, and do it faster and cheaper. AI, of course, seems like a logical step- a tech entity that can work on analyzing data lightning fast, around the clock, without taking breaks, oh and with the power to replace actual employees or take over entire departments. Yes, the media is abuzz with this as well. I'm sure managers everywhere salivate at the idea of having an assistant like Jarvis from Iron Man or the computer from Stark Trek, so they can ask it questions and get excellent answers on how to set business policies and strategies- all without paying salaries and benefits. Well, not so fast (unfortunately). True, there are for example some powerful algorithms in place already in the business world- for examples in trading stocks and running marketing campaigns. For example, Facebook's algorithm(s) has come under a lot of scrutiny and received lots of media coverage this year. On the advertiser side, however, I found out that thing is beyond mystical. I even had people tell me they got so frustrated by Facebook's algorithm that they literally went over Facebook's own resources and manuals and found that even they didn't have the answers. For example, early on in a marketing campaign, you get this metric called a Relevance score; it goes up to 10. The problem is, everyone including Facebook has their own idea of how it's determined and what may make it change. Oh, and the difference between a score of 3 and 10 might mean double, triple or more exposure for the same money! Yet, nobody can tell you how to get a 10 every time! Why is that? Well, my marketing friends told me that it's simply because even Facebook itself may not be totally in control over what they created. I was given an analogy of having a key employee in the company that the whole company relies on, yet you can't talk to that employee and ask it to explain his actions or rationale- ever! This is infurating to many small and e-commerce businesses whose existence and livelihood depend heavily on Facebook advertising. Yet, you can't talk to the algorithm, reason with it, or ask it how you two may work together on achieving better results in the future. How is this ok, and how is this acceptable? Also, where's the accountability when you do everything right, and it screws up? Oh, and what about the fact that this entity works for your competitors at the same time?!
So, what is AI at this stage, really? It's prone to errors, it's mute and beyond accountability and reproach, its abilities are widely misunderstood by the general public, and what it can do now will possibly do in the near future have both been blown out of proportion by the media. I think we as a society continue to look for quick fixes and miracles where there are none. We need a more sober approach to this entire field, and we need to treat its baby steps as baby steps. And, on a final note- whenever you use AI for something, have a plan B or hedge your bets! :)
Let's start with cars. To quote Morty, jeez oh man! (only saw a few episodes so far, still working on understanding all the references haha) We have media blaring out how truck drivers will lose their jobs, taxi and delivery drivers too, how uBer is teasing driverless flying taxis, and my oh my it just doesn't stop. I think these were written by people who don't drive (especially in Toronto, the 6th worst place for commuters- IN THE WORLD, and I don't think it's simply because we don't have AI drivers yet). I'm not much of a driver, to be honest, but let's take a quick look back at the first Tesla fatality. The car didn't understand that a speedway with an intersection had a big truck going through it and the car smashed into it at full speed. No healthy, well-adjusted and focused human driver would make that mistake! Sure, I know Tesla says you need to keep the hands on the steering wheel and the car drives itself only under certain conditions. But, can we take a step back and really look at what's being done to us, the public? We are told these cars have an AI self-driving mode for which you need to keep your hands on the steering wheel and it only works under certain conditions but even then not well enough. Would you accept this level of compromise from your current car? How about your current smartphone? No, you wouldn't- and neither would I. Oh, but it's assisting us with our driving. Hmm. If I had an assistant like that who was so limited and prone to fatal errors, I would be bogged down by stress so much that I would be much better off doing the work myself. Besides, why let my own skills for performing certain tasks go bad because I outsourced those tasks to a crappy assistant? Oh, and when we look at alternatives, there isn't much to see. There's uBer, Waymo, Alphabet and other companies competing to make true self-driving cars a reality. Well, the AI can have fun dealing with Toronto flash rains and flooded roads, Nunavut winters and snowed in roads, black ice situations, low visibility, and human errors of other drivers who are still driving their own cars. Finally, do people who work on creating and teaching self-driving AI's actually drive on a daily basis?! If techie stereotype are to be believed, they most likely don't, or don't drive well at all. Simply put, I don't even think that conventional approaches to programming software are going to cut it here. We need something new and incomprehensible in the mix to make this one a reality. Truck drivers of the world can sleep soundly tonight.
Data analysis, whether for marketing, finance or other fields, has been the buzz term everywhere for a long time now. Every business that wants to stay competitive is now hungry for data, as analyzing it allows them to do more things right, and do it faster and cheaper. AI, of course, seems like a logical step- a tech entity that can work on analyzing data lightning fast, around the clock, without taking breaks, oh and with the power to replace actual employees or take over entire departments. Yes, the media is abuzz with this as well. I'm sure managers everywhere salivate at the idea of having an assistant like Jarvis from Iron Man or the computer from Stark Trek, so they can ask it questions and get excellent answers on how to set business policies and strategies- all without paying salaries and benefits. Well, not so fast (unfortunately). True, there are for example some powerful algorithms in place already in the business world- for examples in trading stocks and running marketing campaigns. For example, Facebook's algorithm(s) has come under a lot of scrutiny and received lots of media coverage this year. On the advertiser side, however, I found out that thing is beyond mystical. I even had people tell me they got so frustrated by Facebook's algorithm that they literally went over Facebook's own resources and manuals and found that even they didn't have the answers. For example, early on in a marketing campaign, you get this metric called a Relevance score; it goes up to 10. The problem is, everyone including Facebook has their own idea of how it's determined and what may make it change. Oh, and the difference between a score of 3 and 10 might mean double, triple or more exposure for the same money! Yet, nobody can tell you how to get a 10 every time! Why is that? Well, my marketing friends told me that it's simply because even Facebook itself may not be totally in control over what they created. I was given an analogy of having a key employee in the company that the whole company relies on, yet you can't talk to that employee and ask it to explain his actions or rationale- ever! This is infurating to many small and e-commerce businesses whose existence and livelihood depend heavily on Facebook advertising. Yet, you can't talk to the algorithm, reason with it, or ask it how you two may work together on achieving better results in the future. How is this ok, and how is this acceptable? Also, where's the accountability when you do everything right, and it screws up? Oh, and what about the fact that this entity works for your competitors at the same time?!
So, what is AI at this stage, really? It's prone to errors, it's mute and beyond accountability and reproach, its abilities are widely misunderstood by the general public, and what it can do now will possibly do in the near future have both been blown out of proportion by the media. I think we as a society continue to look for quick fixes and miracles where there are none. We need a more sober approach to this entire field, and we need to treat its baby steps as baby steps. And, on a final note- whenever you use AI for something, have a plan B or hedge your bets! :)