AI Competency Roadmap
I love technology. I love apps. Also I’ll be the first to admit the design and development process is 80 to 92% tedious bordering on nightmarish. And it compounds. An application that reflects its developers frustrations, easily leading to carelessness, turns people off the experience. They might still have to use the app, as it’s increasingly mandatory by members of learning organizations. But they’ll do so with disinterest, if not outright contempt. All of which makes the process of getting quality user feedback more difficult than it already is.
Now, bring in AI. A technological revolution that if yet to find an absolutely dazzling form in both form, function, performance and economic efficiency. While many apps already ship a light version of statistical learning algorithms, very few tools can claim to be “intelligent.” The ones closest to claiming as much are still pretty confusing and most of its users tolerate the “smart” features rather than enjoy it. The ones closest to provide enjoyable experiences are usually made by big tech, with deep pockets and vast data lakes, not very willing to exchange their insight in full.
There is merit for a debate about the social, economic and political implications for user who are subject to algorithms they themselves cannot inspect. They are, in fact, taking place. But focusing on the learning side of things, there are clear problems if AI, as any sufficiently advanced learning technology, is introduced in classrooms, but not on syllabi.
Popular Science AI
As it turns out, many of the companies most vocally aware of the divide between AI technologists and users are in China. Huawei, Baidu, Tencent, SenseTime and ByteDance (the one behind TikTok) have made a sport out of announcing new features in social media, with millions of users being the judge. This adds a growing pressure to create technology that everyone can somehow appreciate, use or talk about. Formidable government support, along with close-knit ecosystems such as Beijing and Shenzhen, with the largest EdTech unicorns on the planet, mean the pressures to add a learning layer to the local AI races are increasingly high.
To be sure, the U.S. has also proven to be a dynamic space in terms of innovation and capital flows. But for a variety or reasons also worth discussing, just not here, the narratives behind the rise of say Silicon Valley or New York AI are just not as close to the common user, let alone students. Interestingly enough, U.S. VC firms seem more willing to invest on user-facing AI on Chinese firms. A fact that at this point can only be considered anecdotal.
The two largest public EdTech companies, New Oriental and TAL Education, happen to be listed on the U.S. stock exchange. They are of Chinese origin. And both have audacious plan to invest in AI throughout the coming decade, either by their own R&D or through acqui-hires. Which may not mean unequivocally that AI is a relentless focus. It is not their only area of investment, and may end up in the space for lack of better high-risk alternatives. Among other things because they have more cash than they know what to do with it, but unlike Apple or Warren Buffet, it's not staying idle nor seeking refuge from a suspected upcoming recession.
Naturally, not every success story from the Chinese environment is a positive lesson. The growing competition is already putting a dent on the margins of AI-infused businesses, which should have been outstanding. Furthermore, there is some evidence that the fast AI cycle is not having a direct effect on brand valuation. If companies conclude that user-facing AI is not good for the bottom line, it is unlikely that AI development halts, but it might no longer cater to end users as directly as it has thus far. The Chinese government might still insist in its AI-infused educational revolution, but by removing the market dynamics, the rest of the world can no longer reproduce the models on their own contexts.
For lack of better open, user-facing AI advice: Popular AI Science?
So how does that leave an average entrepreneur interested in adopting AI towards new and groundbreaking learning experiences? There are some possible avenues, but admittedly, real evidence is hard to come by at this point. An analysis of the AI ecosystem could give you some preliminary pointers towards deeper research and outcomes.
The takeaway remains the same, AI or not: Focus on your users' problems. Learn more, learn better, learn faster.
AI on the layers: Application, Technology, Foundation
From a user-centered perspective, be encouraged to link as directly as possible developments in the fundamentals of AI innovation. Both in China and the West, the largest companies are present throughout the layers, investing on basic research as well as awareness. This can, arguable, become an opportunity for those who find gaps on the trajectory of a given company. In fact many have been acquired for this reason.
Companies on either side of the world are making efforts to bring their innovations into the other half. Political squabble notwithstanding, the result has been clear, as local players respond with further innovation. One example are autonomous cars. Still, another opportunity lies, in the ability to grasp what AI is solving somewhere, and translate it somewhere else. In fact, AI itself has often been a key ally of internationalization. A case in point: Machine translation has been scientifically linked with a sizeable increase volume of international trade.
The classroom roadblock
Expect TAL and New Oriental to launch brand new AI-infused services for their students in the coming months. But hold tight and don't be surprised if the results are not impressive. Almost all large companies involved in user-facing AI realize it will take a long time for real breakthroughs to really put AI on the hands of everyone in a way that makes sense and has an actual impact. R&D will continue to get a higher piece of the pie than UX.
To end, there is an interesting dilemma for these companies. Up until now, the progress of AI has been largely linked to a large and vast of high-quality (and many here would also add, diverse) dataset. Assuming this is the case, a limited focus on user experience, especifically the interface ability to obtain data and feedback, can compromise a company's rate of progress. But, a small but loud faction of the AI development community offers a stern warning: There is a possibility that the reliance on data-driven R&D is compromising avenues of research with a higher potential in the long term. That would be things like learning efficiency as it relates to the volume of data; or the very issue with which we started: A new stage in engagement technologies that reveal new depths into the user mind and behavior. Yes, this would lead to the need to capture more data, but only after radical new insight about the figures that matter most for the learning process.