The integration of artificial intelligence (AI) into, well, everything, is often a story of promise; the promise of a simpler experience at work with better information to hand and none of the boring administrative tasks to slow you down; the promise of fast customer support, of saving time and money by using bots to automate processes.
The huge leap forward in generative AI we’ve witnessed since late 2022 has added fuel to the fire — not only mainstreaming its use but also equipping bots with more smarts than before.
A recent example: Swedish fintech firm Klarna announced in February that its AI assistant — powered by OpenAI — was handling the work of about 700 full-time agents, and that its chatbot was handling about two-thirds of Klarna’s monthly customer service chats. That’s 2.3-million customer service chats in 35 languages being processed using its on-app chatbot, specifically to help customers around processes like returns and refunds.
Klarna has been open about its use of bots, and its intention to reduce headcount in customer services as a result of these successes. It acknowledges the job threat this represents, and how that will affect individuals. Still, CEO Sebastian Siemiatkowsk tweeted: “We decided to share these statistics to raise the awareness and encourage a proactive approach to the topic of AI.”
You can hear the management consultants salivating from here, but there is definitely something to be said for this transparent approach. We know AI will change the workplace, and many believe it will create more jobs than it will eliminate, but that is a big-picture statement that tends to obscure the pain of change in the short term.
The efficiency and cost-savings versus labour debate rages on, as does the question of how to manage AI’s use in political disinformation, but there is a more subtle threat to add to the mix: the rise of zombie users on social media and how bot-to-bot engagement could undermine the value of social.
A social bot is one that imitates human behaviour on social media platforms by interacting with content and human users — posting, sharing, messaging and liking. We know bots are being used by companies and individuals to shape and shift engagement on social media, for various reasons that extend far beyond influencing your vote — though that is part of it. The concern is what happens when bot engagement overwhelms the human kind.
We know social influencers have been able to buy fake followers for a while, and the bot bonanza has kicked that into hyperdrive. And they’re far harder to spot when their profile pictures aren’t pilfered (and therefore easy to find using reverse image search) but created by an AI image generator, as are their pithy comments and liberal likes.
This isn’t merely about whether Joe Bloggs in your DMs is real. It is also a risk factor for business. Take those using digital marketing, which is arguably almost everyone. Social media isn’t the add-on to a marketing campaign it used to be; it is integral. If the balance tips the wrong way, social bots deployed poorly by marketers become a reputational threat, and third-party bots engaging with marketers’ efforts will distort campaign analytics.
Then there is the value we collectively place on social media engagement. If you see 50,000 likes on a post on your Facebook feed, you might think it was worthy of your time and attention. But if we knew a significant portion of those likes came from bots, most of us would instinctively value that engagement less. And — spoiler alert — with a lot of the most viral content, they do. The problem is estimating the proportion.
Social media, like Meta’s Facebook and Instagram, and Elon Musk’s X, often report on the number of bots and fake profiles they remove, which number in the millions and billions. New York Times tech writer Jack Nicas wrote in 2021 — before the ChatGPT watershed moment, please note — “Facebook said it blocked 4.5-billion accounts in the first nine months of the year, and that it caught more than 99% of those accounts before users could flag them. That number of accounts — equivalent to nearly 60% of the world’s population — is mind-boggling.” But, as Nicas argues, a more significant problem are the fakes the social platforms don’t spot, or don’t spot before they’ve had direct influence on millions of users.
There are companies that have launched on the provision of these bots, but advances in AI mean you don’t need to be a coder (or a company of them) to create a simple bot or to automate the publishing of AI content on your profile.
It’s a boon right up until users — the human ones — realise they’ve been duped. The fake content scandal at Sports Illustrated is a perfect example of this. We’ve also seen this play out in online multiplayer gaming, where not only have players deployed bots to cheat or “assist” them, but developers have used bots as fake players to overcome our natural reluctance to enter an empty game, or to pad out teams.
Getting on top of the problem is not only increasingly difficult, but, in some instances, counter to a social network’s own interests. Of course, a Facebook feed overwhelmed by obvious bots would offer reduced utility to real world users, so Meta must keep this dystopia at bay. Then again, we have plenty of proof that engagement breeds engagement. The mere fact that a post or video served to a user is “trending” presents a lure to engagement. People want to see what the fuss is about, and don’t want to miss out. Engagement is how these firms sell advertising. That’s how social has eaten print media’s lunch.
If the advertisers themselves aren’t reaching real buyers, will they continue to funnel their funds to social platforms? If all your interactions are bottified, will you find new platforms? The zombots are coming, and they might find all the real brains are gone.
• Thompson Davy, a freelance journalist, is an impactAFRICA fellow and WanaData member.
Would you like to comment on this article?
Sign up (it's quick and free) or sign in now.
Please read our Comment Policy before commenting.