The AI data apocalypse
With the rise of Artificial Intelligence, will companies who have the most data gain control of the world?
Artificial Intelligence is one of those elusive tech concepts that we’ve been hearing about for the majority of our lives. Scientists and companies constantly come out saying ‘Now we’ll bring you real AI’…
….and then they don’t.
But these days, Artificial Intelligence is no longer a thing from The Matrix, it’s a part of our lives. If you think that there is no AI in your life, then you’re most likely wrong. Task-specific AI has been with us for a while now. What we’re still waiting for is Artificial General Intelligence.
Artificial General Intelligence (AGI) will be the type of Artificial Intelligence that closely mimics human intelligence and, over time, can surpass it. It will give rise to the ‘thinking machines’, just like in the movie Her.
While we don’t have AGI just yet, we do have Artificial Narrow Intelligence (ANI) — task-specific AI, which analyses vast amounts of data and identifies patterns based on it, making it superior to humans in many different tasks. It is estimated that there are currently over 5,000 ANI algorithms at work in technology companies around the world, with many more to come over the next few years. Basically, every time you perform a Google search that gives you suggestions, ask Siri or Google Assistant about the weather, order something using Echo, or receive a suggestion from Amazon about items you might be interested in, you’re actually engaging with an Artificial Intelligence algorithm.
Such Artificial Intelligence makes our lives easier (except when Amazon desperately wants to sell you a bunch of potty-training products and you’re a 23-year-old student!), and we generally enjoy using it. It’s nice, and contained, and we don’t worry that Siri will take over our iPhones. But AI is gradually heading towards AGI, where even its makers don’t exactly understand how it will behave. A few months ago we found out that Google Translate, the tool that a few years ago was near enough useless at translating anything more than short phrases, has invented its own language. Instead of using English as an intermediary language, as it was taught to do, it started translating between languages directly — which it was definitely not taught how to do — and Google has no clue how it did it.
Why are we so blissfully unaware of how AI actually works? Let’s go back to when DeepMind’s AlphaGo became a victor at the ancient game of Go. To achieve this feat (which researchers didn’t expect to happen for at least another decade), AlphaGo played 150,000 games (so more than any human in their lifetime), automatically making billions of tiny adjustments to its technique. Let’s face it, no human mind can possibly create, or even follow, an algorithm of this size. AlphaGo thus used the data it was provided to learn by itself.
And that’s where we come to what we’re here about — the data. Where does data fit in with Artificial Intelligence? How are they connected?
For ANI, there is no other way to learn than through analysing tremendous amounts of data, just like AlphaGo. The Google search’s smarts didn’t come from a bunch of humans coming up with possible search queries — it learnt (and still learns) from what people were actually searching for. And this basically means that Google is getting smarter by feeding on your data. Data which, with every search you conduct, you basically hand over to them for free, allowing Google to improve its search and build a detailed profile of your internet activity — and you as a person. Some AIs are trained using data you willingly provide (like Google Search), and some with data you didn’t even know was shared, like in the deal between DeepMind and London Royal Free Hospital.
At the moment, there is no AI that can function without vast amounts of data, the same way as there are no surgeons permitted to operate without extensive surgical training (well, let’s hope so!). This leads us to a simple conclusion — companies that have the most data available to them will have the upper hand in the development of AI. And they’re making use of it, with more and more tech companies dabbling in the AI world. The problem with that? Theoretically — none. Capable companies using their resources to advance the technological innovation.
In reality? We might be headed towards a future where those who now control the data, will control the world. Sounds rather dystopian and far-fetched, but let’s look at where the world is now.
An NGO called Global Justice Now has prepared a list of the world’s biggest economies, including both companies and countries, by annual revenue. And if you expect it to be a neat list of countries, then you are sadly mistaken. In the top 100, you’ll find a mere 31 countries. The rest are large corporations. Which means that Apple has more revenue than Belgium, Switzerland or Norway, with the cash it has at hand exceeding the GDPs of ⅔ of world countries. And that’s just one company.
To illustrate this point, the map below compares the ‘net worth of companies’ to the entire stock market of a country.
But let’s go back to tech companies with AI-developing capability. 10 years ago, the tech giants of today were in their infancy. Today, 5 of the 10 biggest listed companies by market capitalisation are in technology. Unsurprisingly, those 5 are Apple, Alphabet (Google), Microsoft, Amazon and Facebook. All of which, funnily enough, are working to develop AI right this minute. And all of those companies derive a large portion of their value from the data they hold. As you can imagine, Facebook doesn’t exactly have billions of dollars of tangible assets. Those new companies are also growing faster than we could have imagined — it took Alibaba just 9 months to reach $100bn in assets. As a comparison, it took the largest money market funds 10 years to achieve the same goal. Similarly, the average value of the S&P 500 index has increased by 16.61% over the past year, while its tech sector portion (which includes companies such as Alphabet, Facebook, Intel, Microsoft and Apple) enjoyed a 35.42% value increase, with most of the previously referenced companies achieving growth in double digits.
It’s daunting to think about the influence these companies already have on our everyday lives. This blog post was written in Google Docs, on an Apple MacBook (which has some Intel chips inside), after conducting tens of Google Searches and, admittedly, looking at my iPhone, checking Facebook, Instagram and WhatsApp a few too many times. When I get home, I will probably relax with some Netflix, or read a book on my Amazon Kindle. Tech is not just a part of my life. It’s my best friend.
This leads us to a question: In a shifting world of globalisation, will the big multinational corporations become the sovereign states of the future? As they grow in size and strength, develop technologies that won’t just change lives of individual people, but whole industries and economies, will anyone be able to control them, or will they turn into city-states of the third millennium?
What constitutes a sovereign state, though? According to Michael Ross Fowler and Julie Marie Bunck a sovereign state needs to have territory, people and government. A small size is not a valid objection to sovereign statehood. It also needs to demonstrate internal supremacy and external independence (but not necessarily an absolute one, as the international community rarely allows countries to ‘roam freely’).
Now, why shouldn’t large corporations meet those criteria? They are governed by boards of directors and C-suite executives, they have tens of thousands of employees worldwide (some of which have even lived in the office), millions of devoted fans, and their offices are large and increasingly self-sustained (the new Apple Park will derive most of its energy from solar panels on its roof). Let’s also not forget virtual realities and their digital territories.
As for external independence? You could argue that those tech giants still have to adhere to the law, rather than make it. But do they, really? Google paid a mere £130m in tax for a 10 year period in the UK and Apple refused to unlock an iPhone after an FBI request. It’s not so clear cut. And the more power they gain, the less likely any country will be able to control them, making them sovereign city-states (or maybe company-states?), with more power over the hearts and minds of the world population than any single country has ever held.
But how will they get there? And how does the data they have at hand play into that?
It all boils down to the data and their Artificial Intelligence capabilities. But not the crippled, context-specific AI of today. We’re talking AGI, the AI of the future.
However, AGI is not going to just appear out of thin air. No one will magically sit down at their laptop one day to create software as smart as humans. Instead, it’ll gradually develop over time, when the current ANI grows its skills, and matures. The companies in the best position to develop the best Artificial Intelligence are those that have access to crazy amounts of data they can feed to it. That data has to come from somewhere, and in most cases, it will come from average internet users, people like you and me. We have absolutely no control over this data — companies collect it without us even realising it most of the time, and that’s the big problem. If this data comes from us, shouldn’t we at least get a say in how it’s used, and by who? At the moment, we are helping big companies gain tremendous control over the world, and getting nothing out of it.
Now is also the last moment to change it. ANI needs a lot of data to learn, AGI will need hardly any, just like a human. By the time we reach the AGI stage, we will already be out of control and, very possibly, out of jobs. As Stephen Hawking has warned, artificial intelligence of the future will hit the middle classes, and hard. If you think that only factory jobs are at risk, you’re thinking last century. Any routine element of your job will soon be automated. In fact, 47% of US jobs are at risk, 35% in the UK, and a whopping 77% in China. And, ironically, it’ll be the data we provide ourselves that will kick us out of our comfy office chairs. Growth in the big data industry is reaching a plateau, meaning that companies very nearly have enough data to make us, the providers, obsolete. By 2018, Gartner predicts that 45% of companies will employ more machines than people. Pretty damn scary.
You can’t stop innovation. People tried to stop steam engines, and factory automation. But that’s just not how progress works. AGI will happen, whether we’re ready for it or not (and we’re not, since it takes an average person 6 years to adapt to new innovation). But what we can do is control how it happens.
Otherwise the big companies will make lemonade from your data, and taking back what’s left of your lemons will not undo it.
Regaining control of our data is one of the most important challenges we face. It’s deciding who will control the world in the future, and what the future of employment will look like. Which is kind of important — we need money to buy new iPhones, after all. But how do you regain control of your data? people.io is here to help you do it. With our app you’ll be able to control your relationships with brands and how your data is used, all while getting rewarded.