Posted on

Using Futures Thinking To Navigate Disruptive Shifts

After the release of the world’s first mass-market ‘personal navigation device’ in 2004, TomTom’s stock soared, reaching a peak of $65/share in 2007. However, the landscape shifted dramatically when Google launched its Maps app in 2007, effectively transforming every smartphone into a personal navigation device. Within 2 years, TomTom’s stock fell below $3. While TomTom was eventually able to pivot and change their business model, their initial dismissal of the threat posed by the rise of smartphone based navigation apps was very costly. This failure was not due to a lack of technological capability but rather a lack of strategic foresight to anticipate and adapt to disruptive technology.

Innovation is not just about responding to current needs but also anticipating and preparing for disruptive shifts, like the rise of generative AI. However, preparing for disruptive shifts is a very tricky problem because the mental models we use to make decisions in normal, predictive times don’t work in disruptive times. Product planning in such times has to go beyond predicting the future (which often turns out to be incorrect), to planning for multiple plausible scenarios. This is where futures thinking can be a useful tool. 

Futures thinking is a strategic approach to uncovering multiple possibilities, with the aim of creating a preferred future. It ties closely to innovation – once we identify a desirable (and plausible) future, we have a clearer roadmap of the problems we need to solve in order to reach that future. 

So how do you develop futures thinking? It starts with first figuring out (or even deciding) a vision for the future and then understanding what forces enable or thwart that vision. Depending on how likely and how important certain trends for a particular vision, one can arrive at plausible scenarios of the future. These scenarios can then guide what kinds of ideas and products to invest in. 

Future Archetypes

When it comes to our vision of the future, we all hold one or more of common archetypes which  dominate our imagined future thinking. Below are the five common future archetypes viewed through the lens of AI:

  • Progress: A tech-driven world with humans at the center, emphasizing rationality and innovation. AI in this future enhances human productivity, creativity, and decision-making.  
  • Collapse: A darker vision where AI exacerbates inequality, destabilizes jobs, and concentrates power, pushing society to a breaking point.
  • Gaia: A partnership-driven future where AI helps repair damage to the planet and fosters inclusive, harmonious systems between humans, nature, and technology.
  • Globalism: A borderless, interconnected future where AI powers collaboration across economies and cultures, breaking down barriers to knowledge, trade, and innovation.
  • Back to the Future: Nostalgia for simpler times, where AI’s rapid advancements are rejected in favor of human-centered, low-tech solutions to protect societal stability.

Trends

There are several key market, technology and social trends that impact the development of AI in both positive and negative ways. Here are a few sample trends:

  • Technology Improvements: The AI hardware market, encompassing GPUs and specialized AI accelerators, is projected to grow significantly indicating growth of computational power. AI models are expected to continue improving with more enhanced reasoning skills and capabilities. 
  • Regulation Focus: The number of AI-related regulations in the U.S. has risen significantly, with 25 AI-related regulations in 2023, up from just one in 2016, reflecting a growing focus on responsible AI development.
  • Computational Costs: Training large AI models is resource-intensive, requiring significant computational power, energy, and financial investment. OpenAI’s GPT-4 used an estimated $78 million worth of compute to train, while Google’s Gemini Ultra cost $191 million for compute.
  • Public Trust: People and companies may be hesitant to adopt AI due to various reasons like biased algorithms, privacy concerns, or fear of widespread job losses. 
  • AI Investment: Despite a decline in overall AI private investment last year, funding for generative AI surged, nearly octupling from 2022 to reach $25.2 billion, highlighting increased investment in AI technologies.

Each of these trends can either contribute to a future or become a barrier. The likelihood of each trend plays a part in which of the futures are more plausible. 

Future Scenario Planning

To determine plausible future scenarios, leaders need to evaluate trends against each vision of the future. For example, consider the future archetype of “progress” where AI leads to greater productivity and innovation. Improvements in technology, like better models or better GPUs, clearly push us towards this future. However, issues like algorithmic biases or security concerns can erode trust in the technology and slow down adoption rates. If this is a preferred vision of the future, then actively addressing these barriers during the development process can ensure that we keep marching in the right direction. While this was an overly simplified example, a more thorough analysis that incorporates additional relevant trends can start to reveal the plausibility of different scenarios.  

One challenge in future scenario planning is that given the complex nature of the problem, there is no good way to accurately determine the likelihood of each trend and its contribution to each future archetype. This is where swarm intelligence might be useful. Groups of people are often better at predicting than relying on experts. Training employees on futures thinking and tapping their individual unique insights might provide better signals on what scenarios are more likely to happen in the future. 

As artificial intelligence redefines industries, businesses must integrate strategic foresight into their innovation frameworks to thrive in uncertain and fast-changing landscapes. By explicitly using future archetypes and integrating them with current and expected trends, we can start to identify what scenarios are most likely to play out in the future. These scenarios can then help create more efficient product innovation roadmaps.  

Posted on

How Technology Can Improve Deep Learning

In an experiment to evaluate the impact of media on learning, researchers showed volunteers a presentation about the country Mali. Some of the subjects saw a text-only version of the presentation while the others saw a multimedia version that included additional audio-visual content.

After the presentation, the researchers gave all subjects a quiz on the material. The text-only group were able to answer more questions correctly on the quiz compared to the multimedia group. The outcome of this experiment was summarized as, “The text-only readers found it to be more interesting, more educational, more understandable, and more enjoyable than did the multimedia viewers, and the multimedia viewers were much more likely to agree with the statement ‘I did not learn anything from this presentation’ than were the text-only readers.”

Technology has undoubtedly made a big impact in education. Apps and games that teach specific reading and math skills have shown to improve learning outcomes, and productivity apps have made research and collaboration so much more easier in the classroom.

However, technology doesn’t just provide us with tools to learn specific skills or be productive, it also actively changes the way we think and process information.

And quite often, these changes inadvertently end up being detrimental to learning in some ways. Professor Patricia Greenfield explains, “Although the visual capabilities of television, video games, and the Internet may develop impressive visual intelligence, the cost seems to be deep processing: mindful knowledge acquisition, inductive analysis, critical thinking, imagination, and reflection.”

Inappropriate or overuse of technology can significantly impair learning, by breaking attention and interrupting the learning process. Our brains contain two types of memory – short-term and long-term. Long-term memory, which can hold information for a long periods of time, is the seat of understanding where complex schemas and patterns that give us meaning are held. Short term memory on the other hand is fragile – it can hold information for only a few seconds. One type of short term memory, called the working memory, is what we use when we have to retain partial results as we work through a math problem or follow a sequence of steps. However, working memory, unlike long-term memory, is small and can hold only a few chunks of information at a time. After the contents of the working memory are processed, they can be encoded in long term memory for future retrieval.

The challenge with this learning process is that since working memory can retain information for only a few seconds (~20 sec), and any distractions in that time interrupt the flow of information to long-term memory. Being able to focus and reflect on concepts for extended periods of time are critical to learning new things.

In addition to inferior learning, poorly designed technology can have other harmful effects.  When the ability to focus on tasks decline, it can lead to feelings of boredom and an increased desire to seek more external stimuli. Time spent with media (television, video games) has been shown to result in ADHD like behavior.

If we want to promote critical and creative thinking, essential for deep learning, we have to unlearn the way technology is designed. Here are some things to pay attention to when designing technology products for use in education that can promote deep learning.

Pay Attention To Passive Switches

Switches are interruptions that result in students switching between different tasks. Passive switches, as opposed to active switches, are those that students don’t initiate themselves. Obvious examples of passive switches are email notifications, chat features, or pop-ups within an app that are meant to help students but inadvertently break their focus.

Less obvious examples of passive switches include using hyperlinks in the text, often with the good intention of providing information to fill the gaps. Unfortunately, hyperlinks also subtly nudge students into clicking before they have had sufficient time to process information, thereby breaking their flow. In one experiment, groups of people were asked to read the same piece online writing with different number of hyperlinks. Results showed that as the number of hyperlinks increased, reading comprehension went down. The researcher explained her findings as, “Reading and comprehension require establishing relationships between concepts, drawing inferences, activating prior knowledge, and synthesizing main ideas. Disorientation or cognitive overload may thus interfere with cognitive activities of reading and comprehension.”

Be Less Helpful

In an interesting experiment, researchers gave students a tricky puzzle to solve that involved moving colored balls between boxes based on some rules. One group of students got a helpful version of the software that had on-screen assistance and other cues, while the other group got a bare-bones version with no hints or guidance.

In the early stages, the helpful group outperformed the bare-bones group in how fast they solved the puzzle. However, as the test progressed the bare-bones group got more proficient and was able to solve faster with fewer incorrect moves as compared to the helpful group, which gave clear indication that they were planning ahead and using strategy.

It didn’t just end there. Eight months after running the experiment, the researchers invited the students again and gave them similar puzzles to solve. The group that used the unhelpful version of the software was able to solve the puzzles twice as fast compared to the helpful group.

When help is too easily available, it robs students of the opportunity to think for themselves and build critical and creative thinking skills.

Be Judicious With Media And Visuals

Unnecessary media usage can overload working memory making it harder to process and assimilate knowledge.

In an experiment conducted on college students, researchers showed groups of students a typical CNN broadcast. One group saw the broadcast along with infographics that flashed on-screen and text-crawls on the bottom. The other group saw the simpler version of the same broadcast without any additional infographics or text-crawls. Subsequent testing showed that the multimedia group retained far fewer facts about the news compared to the simpler group. The researchers theorized that the “multimessage format exceeded viewers’ attentional capacity.”

Keeping things simple when working with different forms of media works much better from a learning perspective. While different forms of media are good to use individually, using them simultaneously can overwhelm working memory.  

 

To design educational technology we need to carefully assess if the technology or feature encourages students to think and reflect, or does it distract them. When we introduced a team related feature not too long ago, we realized it was working a little too well, to the point of getting in the way of real learning. We decided to remove the feature and will likely introduce it again in a different incarnation, where it improves productivity without being a distraction.

Technology has great potential to improve student learning in different ways, but it requires us to be more mindful of the learning process while designing it.

 

Posted on

Designing Products to Build Intrinsic Motivation

In a recent study researchers wanted to explore the relationship between rewards and motivation in the context of education. In order to understand the impact of gamified elements on student motivation and learning, they designed a long-term study for students enrolled in a semester long course. Students were divided into two groups – a gamified group that used a reward system aligned with the learning goals, and the control group that received the same instruction but without any gamified elements. They looked at student grades at the end of the course along with student surveys, and confirmed what some educators had always suspected.

The researchers found that the non-gamified group not only did better at the end of the semester exam, they also reported higher levels of motivation and satisfaction at the end of the class! As the researchers explain, “The results suggest that at best, our combination of leaderboards, badges, and competition mechanics do not improve educational outcomes and at worst can harm motivation, satisfaction, and empowerment. Further, in decreasing intrinsic motivation, it can affect students’ final exam scores.”

While typical gaming elements like points and badges can lead to increased engagement in the short term, it is now believed that the initial appeal is due to a novelty effect, and that engagement and motivation decline as the novelty wears off. And this effect is more pronounced for younger age groups, where novelty and interest declines faster.

Educational products routinely employ rewards like badges and scores to get initial interest and traction among users, however, as research is now pointing out, these elements have negative long term consequences as they promote extrinsic motivation instead of building intrinsic motivation among students.

So,  how can we design educational products that focus on building students’ intrinsic motivation?

Edward Deci and Richard Ryan, professors of Psychology, have studied motivation for several decades and developed the Self Determination Theory (SDT) of motivation. According to their theory, three innate psychological needs play a role in motivation – competence, autonomy and relatedness. The main premise behind their theory is that humans have an inherent tendency to learn, have agency in their development and connect to others. Their theory has been widely used in many contexts, including gamification.

Based on the underlying theory of self determination, here are some high level product approaches that can be used in lieu of rewards to build the right kind of motivation:

Exploration

Creating a playful environment that leads to self-directed exploration ties to the underlying need for autonomy and competence. Games or products should allow for the freedom to fail, by allowing users to recover from mistakes without penalty. Games should also provide a freedom of choice, where users can decide what they want to work on or what skill to develop.

Feedback

In a classroom, feedback can be slow and constrained as teachers can only provide feedback one at a time. Games where feedback can be immediate can have a positive impact on the need for competency. Feedback messages that are actionable (guide the student in the right direction) and focus on growth mindset have been found to be effective.

Collaboration

A typical classroom environment fosters competition among students instead of collaboration, which in turn reduces intrinsic motivation. Elements like leaderboards have the same effect due to social comparison. A better way would be to design products that allow meaningful collaboration among students, and tap into the need for relatedness. Social cues that signal working together have been found to boost intrinsic motivation.  

 

Intrinsic motivation has been found to link positively to learning outcomes as well as personal wellbeing. Introducing the right kind of gamified elements into product elements can boost intrinsic motivation among students, but it involves walking away from more traditional elements in games like badges and points.