The internet, as it stands, is a global system of interconnected computer networks that allows humans to interact with each other, do business, gain knowledge, and be entertained. Google allowed us to find everything we need. Amazon allowed us to buy everything we need (or don’t need). Facebook, Twitter, and Instagram allow us to share everything about ourselves for others to see.
Right now, we spend plenty of time at our screens clicking, chatting, sharing, and searching. But how much longer will that last? The internet could soon become a very lonely place, as the rise of bots handling tasks for us online may lead to a dramatic shift in how we use the web.
Stage 1 (Already Happened)
In the beginning of the internet, we had to do everything on our own: Ping the right machine to transmit information to the right person. Find the information we were looking for. Organize it. And so on.
Fast forward a few years, and service providers like Google started to “help us sort” our search results. Mail apps put certain things into a spam folder, and Facebook “helps us sort” what it thinks we should see. This is already true in today’s online interactions. But now…
Stage 2 (The Bot Environment)
The sorting was only the beginning. Thanks to machine learning, artificial intelligence, and the bots that are created to do certain things for us, online interaction has already shifted quickly to being heavily human-bot. Bots are everywhere in today’s online world — they chat with us on shopping sites, perform customer service at all kinds of call centers, arrange appointments for us, and even influence elections using social media.
On top of all this, many interactions online are now also evaluated by bots. For example, bots are analyzing job or loan applications and automatically replying if the analysis shows an applicant is not the right one. Bots also send us recommendations of how we should change our resumes or loan applications to pass the bot on the other side (you can even pay the bot to do that for you).
We are talking to bots, reading what bots post, arranging meetings with bots, and chatting with bots—often without even realizing it, since they have names like Amy, Susan, or Matt. And bots like re:scam now also engage with spammers to keep them busy (which actually is kind of funny).
Stage 3 (Bot vs. Bot)
It’s clear that companies have identified the value of bots and integrated them into their sales, customer service, and human resource processes. But now individuals are also starting to use them to navigate through the bot-heavy world of the internet, which leads to bot-bot interactions where humans are not involved anymore.
Let’s go back to the example of applying for a job. Companies already use bots to scan, filter, and interact. But now users can also hire their personal bots to find jobs and apply for them using the information the applicant provided upfront, with the bot creating the cover letters and even applying or replying. Or to take a simpler, already very common example: Personal bots now arrange appointments with another person’s bots—and maybe we both show up, or maybe we don’t.
Here’s where the problem starts.
Where This Leads
The problem is simple. If bots are now more efficient than people, they can get things done in a way humans can’t — and we all know this. But when there’s also a bot on the other side, things start to get out of control.
If my bot applies for 100 jobs or loans and there are 100 bots on the other side replying, the interactions grow exponential. This is true for chats, mails, posts, likes, tweets, snaps, and every other online interaction you can possibly imagine. The sheer quantity that grows from these simulated interactions means humans themselves won’t be able to consume, filter, read, and reply anymore.
It’s just impossible due to the current limitations of the human mind. In the best case, this means people will get a “sorted” result of what a bot thinks we should see, missing 99 things that could have also been interesting. The paradox is that this bot glut could eventually push most human interaction offline again; news (real news, that is) will be shared by talking, jobs will be found through connections, and friends will discover major life updates about one another at events and reunions.
This is the best case. Another option is that we will have bot-free zones online.
The problems that come with letting this scenario worsen—that is, giving bots more and more control over our everyday online activity—are dramatic. Starting with the computation power that will go into these interactions, the electricity used by the machines, the data created, and the storage space needed for the data. If this happens and grows exponentially, we might not have SkyNet (yet), but 99 percent of online interactions will not be human anymore, and machines will continue to slowly take over the internet—not to mention the phone lines, as we recently witnessed at the Alphabet keynote.
With AIs that can talk to people on the phone, we’ll see an increase in AI on the other side as small business owners start introducing digital counterparts in the form of bots as a service (Baas). Once AIs start talking to each other, they might start talking in ways we don’t understand or about things we don’t want them to talk about. It’s happened before; in 2016 Microsoft had to shut down its chatbot Tay when it began to post offensive tweets, and last year Facebook shut down a project in which two AIs seemed to be chatting to each other in a language only they understood.
As more components of our lives become automated, we may want to give some extra thought to which of our routine human interactions are ok to reduce to a bot, and which are worth doing the old-fashioned way, with our own voices, hands, and eyes. If the human vs. bot world we’re already experiencing seems overwhelming, the bot vs. bot world will only grow more tangled; in trying to pass our daily tasks on to computers, we could end up creating a lot more work for ourselves than we intended.