It’s almost certain that we all know friends and co-workers who take different points of view on digital adoption. At one end of that continuum, you’ll get the “if it ain’t broke, don’t fix it” brigade, who would probably still be using Windows 98 if they could. These are the people who see digital adoption as an unnecessary evil, who are usually brow-beaten into accepting software updates and software innovations at work and at home.
Then there are those in the middle of the digital adoption road, who just set their computers to ‘auto update’, couldn’t care less what happens in the background, so long as everything works, and they’ll ask a friend or colleague for help if they get stuck.
At the other end of the spectrum are the early adopters, who welcome digital change with open arms, often excited to see what new software or upgrades will bring to their daily lives.
Unsurprisingly, it’s this latter group who are the most comfortable with artificial intelligence (AI) being used to drive the process of digital adoption, which is accelerating at an exponential rate. The process of advancing technology creating its own momentum of further advancement is linked to a concept known as Moore’s Law, which is the theory that integrated computer circuitry will double in processing speed and efficiency every two years, with little cost increase.
As this efficiency doubles, so does a human’s needs to perform the same tasks ever faster and with more automation. Some call it Moore’s Law – others call it a vicious circle of unnecessary hell!
The Turing test surpassed.
The Turing Test refers to an assessment set up in 1950 by Alan Turing, a genius British computer scientist. In simple terms, it was designed to determine whether a machine could exhibit intelligent behavior that was indistinguishable from human expression.
The test was always conducted by a person typing commands into a computer terminal placed in a separate room to another terminal. That second terminal would either be operated by a computer itself, or by a human replying to messages sent by the tester. A computer would have been deemed to pass the test if the tester could not distinguish replies from a human. In effect, if the tester thought that the responses were coming from a person when in fact they were created by a machine, the machine would pass the test.
With the advent of chatbots and platforms such as Chat GPT, we can now firmly state that the test has been surpassed, but for many years since 1950, that has never happened.
In fact, a study cited elsewhere on this site showed some startling results:
- GPT-4: was mistaken for a human 54% of the time.
- GPT-3.5: was mistaken for a human 50% of the time.
- ELIZA: was mistaken for a human 22% of the time.
Interestingly, real humans were correctly identified only 67% of the time!
AI as a contextual co-pilot.
Weekly AI essentials. Brief, bold, brilliant. Always free. Learn how to use AI tools to their maximum potential and access our AI resources to help you grow. 👇
Now we’ve seen just how smart AI can be, let’s look at how AI tools can assist people with the digital adoption process via automation, analytics, and improved user experience (UX). In fact, according to WalkMe, an advanced digital adoption assistance platform, AI can “analyze, automate, and optimize experiences to eliminate digital friction.”
Digital adoption platforms (DAPs) work by running a ‘teaching layer’ of software alongside the package they are designed to assist. In basic terms, the DAP helps in the form of tooltips and walk-thru videos whenever an employee or software user makes common mistakes. But crucially, the AI learns to predict every individual person’s workflow, only offering assistance where necessary. Once a person has mastered a particular task, the DAP will understand this and move on to helping in another area of the person’s expertise.
Under this system, in a perfect world, once a software operator had been using a DAP for a certain period, it would never need to show its presence again, until software is updated, or a new employee starts at the same terminal.
Natural Language Processing (NLP)
NLP allows people to interact with software using their normal voice and / or by typing, in whatever language of choice. This can be extremely useful for multinational companies where reports can be translated to another language on the fly or Zoom calls could be subtitled with translations in real time.
What’s more, autocorrect and predictive text can help users input data more efficiently and with fewer errors (always provided they have the option to turn the ducking thing off!!)
Personalized onboarding and training
AI can customize onboarding and training processes to individual users’ needs and learning styles via hyper personalization on a user / account basis. For example, as mentioned above, the AI in the digital adoption assistant reacts differently to users dependent on their workflow patterns.
This is known as adaptive learning, whereby the DAP can create interactive, step-by-step guides that help users through new screens or UX, providing immediate feedback and assistance.
Bots can also perform routine tasks such as resetting passwords or walking people through complex software features. But these processes are driven by data analysis; not merely a ‘blanket’ approach.
By tracking how users interact with new software or updates, AI can monitor system performance and predict potential issues, ensuring that tech problems are addressed before they affect users.
Weekly AI essentials. Brief, bold, brilliant. Always free. Learn how to use AI tools to their maximum potential and access our AI resources to help you grow. 👇
This is, in effect, automated troubleshooting, which reduces platform downtime and improves users’ confidence in the tools they use.
User segmentation.
By segmenting users based on their behavior, AI can assign users into groups and report to management via an analytics dashboard. This allows HR departments and trainers to customize further professional development to various groups, rather than effectively teaching mixed ability classes.
A/B testing
A/B testing can be an extremely complex process to analyze successfully. But handling huge amounts of data is just what AI was intended to do. Accordingly, A/B testing results can quickly demonstrate which upgrades or changes are working for employees and which prove more problematic.
A DAP’s AI can run experiments to test different versions of a feature or interface, identifying the most and least effective sections of software.
Fraud detection
AI can detect, predict, and therefore prevent fraudulent activities. For example, someone logging on to a software package from home late at night or adding payees that don’t fulfill certain pre-approved criteria. AI in DAPs can also ensure that ‘special category’ personal data under privacy laws is identified and not displayed, even to management.
In summary
At the beginning of this article, we mentioned the digital laggards who object to tech changes in their lives, yet perhaps they are the group of people who will find the most help from an AI-driven DAP. After all, if new processes can be achieved seamlessly and offer genuine utility to every user, that must be a good thing.
The only caveat to perhaps heed; ‘to err is human’ – so it’s important not to single out and treat digital slowcoaches unfairly. Those who have difficulties with new tech are often the most creative thinkers or practical real-world problem solvers. After all, if you’re introducing changes in the workplace or with established routines, who better to ask for feedback than those who find those changes more challenging?
