Elon Musk, Twitter, and the need for AI literacy

Focus on raising general awareness of the power and ubiquity of algorithms

Perhaps you have been following the ongoing saga of Elon Musk’s quest to purchase Twitter for $44 billion as I have this month. It’s a fascinating storyline full of daily drama featuring a colorful character who apparently is now a fellow Texan and is the richest person in the world (or was last time I checked). There’s been a lot of “will he or won’t he” debate on whether Elon Musk’s bid to acquire Twitter will ultimately go through (it’s “on hold at the moment due to Elon’s concern about the number of bots on the platform). While it is certainly entertaining, sadly I don’t have any special insights on how this will ultimately play out or the potential implications for politics, business, or content moderation on a major social media platform going forward.

Elon’s warning

However, there is one part of this week’s drama featuring Elon Musk that specifically caught my attention: the debate over a setting that Twitter users can select on their Home Screen. (I’ve highlighted part of the below tweet for emphasis.)

For those who aren’t familiar with Twitter’s interface, what Elon is referring to are the “stars” that appear on the top right corner. If you click on them as a user you’ll see two options: Top Tweets or Latest Tweets.

The default setting is Top Tweets. What does this mean? In a nutshell, it means that Twitter is using an advanced, proprietary algorithm to determine what tweets you see in your view as you scroll. Importantly, Top Tweets include not just the accounts you follow, but accounts that Twitter thinks you would be interested. You’ll see commentary on tweets from accounts you don’t follow indicating whether someone you do follow likes the tweet or follows the account, as well as topics that Twitter decides are relevant to you. For example, I see a lot of accounts relating to basketball, San Antonio, and colleges (my son Andrew is a high school junior.) The alternative is selecting the Latest Tweets option, which simply shows you the tweets from the accounts that you follow in reverse chronological order - no artificial intelligence (AI) necessary.

Which choice is best? That depends on your perspective, as nicely encapsulated in these two replies to Elon’s original directive.

In reading through some of the replies to Elon’s original recommendation to switch the default setting from Top Tweets to Latest Tweets, you likely won’t be surprised to hear that the replies ranged from appreciation to skepticism to outright cynicism…both directed at Twitter and Elon Musk! Most people who commented seemed to have strong opinions on the matter…again, not a surprise to anyone who is an active Twitter user. Elon Musk himself is known to have strong views on AI and what he perceives as its negative impact on humanity. However, what caught me off guard was the substantial number of people that were not aware this was a setting in Twitter that could be changed.

There is a profound lack of awareness about algorithms and their daily impact on our lives

I’ll admit that I’ve grown accustomed to the Latest Tweets model that was dominant for the early years of Twitter and never bothered to switch the setting once Top Tweets became the default. There are aspects I like and don’t like of both views, and I suppose inertia won out in my case. However, many people were completely unaware that Twitter was using an algorithm to optimize their experience and that, if they didn’t like what Twitter recommended they see, users had the power to easily change this to the more traditional view of only seeing accounts they follow in reverse chronological order. Yes, in Elon’s parlance, you are being manipulated by the Twitter algorithm in ways you don’t realize. However, I would make a key distinction here between:

1. people who recognize that Twitter is using an algorithm to determine their Top Tweets, just lack specific knowledge of exactly how they are (the algorithm is proprietary, so you can’t know exactly how it works)

2. people who are entirely unaware an algorithm is being used, period.

I’d argue that users fall into the first category are like most people when it comes to interacting with any type of technology. I don’t know how much car or computer works, but I still get value from using them without having any detailed knowledge of how they operate. It is users in the second category that I am more concerned with - they lack what I’ll term “AI literacy”.

This general lack of AI literacy shouldn’t be surprising to me because I’ve seen it in my work for years. One of the stories from my book The End Of Insurance As We Know It which resonated the most with readers was about a road trip I took in 2018 with my elderly father who brought along a bag full of paper maps to help us navigate. My father had many more years of experience traveling the highways that we were traveling on. Yet despite my father’s practical knowledge from previous trips driving on the roads and his trusted paper maps, I had access to superior technology using GPS on my smartphone which could account for factors such as road construction, accidents, and traffic congestion in recommending the fastest route. Our reliance on different tools led to some heated arguments along the way on which route to take, all because my father wasn’t aware that this AI-based technology called GPS existed!

Despite my father’s lack of familiarity, it is remarkable how smoothly for most people the transition has been from using paper maps to GPS on smartphones over the past decade As long as users know how to pull up Google Maps and get driving directions, it doesn’t matter if they understand how the software works. Until confronted with someone who still relies on paper maps, we don’t really think too much about the myriad underlying technologies and use of AI algorithms that make GPS possible. Why not? Because it is unnecessary to have this knowledge, even though the GPS algorithm is manipulating us in ways we don’t realize. This lack of knowledge of how GPS works (most of us) is very different than the lack of knowledge that GPS exists (my father).

How important is AI literacy?

I was reminded of the relative lack of awareness of algorithms in general during a recent training session I conducted on behalf of the Efma Business Academy on AI-driven processes. I’m passionate about the power of automation and AI in the insurance industry, having written articles and co-authored a book called Hyperautomation on the topic. At the beginning of my corporate training session for a range of professionals representing different domain expertise, I polled the group to see how familiar they were with advanced algorithms and asked if they currently used AI in any of their daily processes. This was a trick question: the reality is that all of us interact with AI every day, multiple times a day. Some participants offered up use cases such as spam filters and auto-correct technology. Others referred to process automation technologies such as optical character recognition (OCR), natural language processing (NLP), and natural language understanding (NLU) in solutions such as chatbots. Fraud detection in claims processes was another common user case that was identified. But about a third to one half of the audience said that they did not use AI at all in their roles. Clearly, there is a gap in understanding the power and reach that algorithms and AI play in our daily routines and processes.

The role of advanced algorithms and AI are important for a variety of sectors in the insurance industry and are a hot topic for regulators in the United States and elsewhere in the world. How much awareness and knowledge is required by your employees about the impact and reach of algorithms and AI in their daily work? How AI literate do your customers need to be? There is likely no one-size-fits-all answer to these questions, but it is important that your organization consider these questions and formulate answers to them.

I believe that broad, general awareness of algorithms and AI is important for a few reasons. First, your employees and customers can have more trust and confidence in using systems that employe algorithms and AI and can potentially make tweaks and modifications that improve their experience - they have what’s known as a “locus of control” and will be less likely to fear the unknown (similar to the option of changing what tweets you see). Second, a working knowledge helps both employees and customers to provide helpful suggestions and feedback that can refine products and algorithms, leading to better user experiences. Third, more familiarity and comfort with how algorithms and AI work in a conceptual sense can help combat more nefarious narratives that undercut trust and can lead to repetitional risks. Fourth, more inclusive conversations can help reduce concerns about bias and promote the ethical use of AI.

What is the “AI literacy” of your organization and customers? Do they know how algorithms are used generally and have a comfort level, even if they don’t understand the technical details? Is there a need for more training and broader discussions to raise the level of awareness and understanding of the power and ubiquity of algorithms at your company?