Will Your Next Website Be Designed By A Robot?
Retro predictions of intelligent humanoid robots that can do everything from looking after the kids to walking the dogs haven’t yet come to fruition. But a subtler and smarter form of AI has long been a part of our lives, quietly taking over our worlds from the inside out.
It may not look quite how we expected it to, but we’re well and truly in the midst of an AI revolution. Computers have mastered the most complex and abstract human games like Chess and Go. Billions of people around the world rely on AI-powered devices to wake up on time, make money, pay their bills, and simply get through the day.
And although not as advanced as the cyborgs detailed in Future Shock, there are AI-powered smart nanny bots that help you keep an eye on your kids and pets while you’re out.
We live in an increasingly digital world in which we co-exist with and depend on intelligent computers. Yet, in typical human fashion, as these bots provide several advantages over us — they’re reliable, and don’t require feeding, watering, rest, etc. — many feel threatened by our very own creations and think an all-out mutiny is on its way.
But aside from in the apocalyptic visions of science fiction books and movies, machines are a long way from developing general intelligence and qualities like common sense, curiosity, and abstract reasoning — the things that make us so unique. The reality is, machines are nothing like us, and that’s precisely what makes them so useful.
So, writers, programmers, and most other creatives who work in the digital economy don’t have to worry (too much) about being replaced by a super-intelligent AI. However, you may soon be sharing your workload with a bot or three. This is largely thanks to one emerging subset of AI that’s already reaching a capacity in which it can masquerade as being able to think and understand like a real human. This is machine learning.
Learning to be a better human
Artificial intelligence is defined as the capability of a machine to imitate intelligent human behaviour. As explained by Pedro Domingos, author of The Master Algorithm, it works like any other algorithm: you input certain data, write the algorithm, and out come the desired results.
Machine learning, however, turns this process inside out. First, you input the data and the desired result, and then out comes the algorithm that converts one into the other. It is essentially problem-solving on a massive scale — taking vast quantities of information and analysing it to make predictions, find patterns, and learn the best way to achieve a particular result.
For example, in programming, machine learning may be used to instantly analyse thousands of possibilities and put together a solution that humans may not have previously thought of. This is not creativity as such, but rather a larger, more connected brain that can search more thoroughly and widely than is possible for mere mortals.
We can already see machine learning in action in everything from how we search on Google and buy products on Amazon to how we find new music on Spotify and make real-time language translations. Its power is in how natural and seamless its scripts are — based upon the way the world works and continuously improving itself as it goes along. All without troublesome subjective opinions or inevitable human errors.
Your friendly, local chatbot
One way machine learning is taking off is by improving interactions on a mass scale. Chatbots, voice and text-based conversational interfaces powered by AI, are being used by many businesses to reach and interact with their customers in a more engaging and personal way.
The Whole Foods Market messenger chatbot is a great example. By mix and matching emojis with words, customers can use the bot to browse its store for available products and then it will help find the right recipe to turn them into a meal. Such one-on-one interactions, across multiple platforms and a limitless scale, are particularly useful for businesses as they enable them to gather valuable information about their audience, analyse it, and feed it back in to create even more personalised experiences.
Chatbot-building platforms like ChattyPeople make it possible for anyone to create their own chatbot in minutes. You don’t need any coding experience or to be on Facebook, chatbots work across many platforms including Google Assistant, Slack, WhatsApp, and Amazon Alexa.
Robots that can see the future
Another way AI and machine learning will support and not replace developers has to do with its practical powers of prevention and prediction.
Microsoft exemplifies this with its Security Risk Detection branch that’s currently using AI to help software developers find bugs in their code and other vulnerabilities. The tool works through a method of fuzz testing — asking “what if” questions over and over to determine the root cause of an issue. It’s proving especially useful for companies that build software in-house or use off-the-shelf and open source products.
The video game company Ubisoft recently took prevention a step further with its new tool, Commit Assistant. By feeding ten years worth of code into a machine learning script, the company created a program that flags mistakes in a game’s code before programmers even make them, and offering solutions based on past rectifications. Ubisoft claim catching errors before a game is released can eliminate as much as 70 percent of the cost of debugging, not to mention vastly reduce the time from development to deployment.
A smart new partnership
Google, in response to a lack of top talent in programming, has already created a set of machine-learning scripts that can do certain tasks much quicker than human programmers. What’s more impressive, and somewhat more daunting, however, is that the system, known as AutoML, can already code better programs than the researchers who built it.
For example, AutoML coded a machine-learning system that performs image recognition tasks with a higher level of accuracy than human-created systems. This may sound like the beginning of the end for us, especially when mentioned alongside chatty AIs and future-seeing programs. But when you think about it, if there’s one thing machine learning AI is going to be able to do better than us, it’s building AI.
As AI advances, it will be put to work finding novel solutions that humans are unable to see and that lay outside our processing capacity. In this way, businesses and programmers will be freed to spend more time on tasks that require the unique and complex qualities of intellect, and the products we jointly produce will be infinitely more intelligent. Even if it is a robot that’s the real brains behind them.