hat gen AI will increase the productivity of knowledge workers significantly is a base assumption. I’m on-board – it’s exciting, and the potential is incredible. But it’s worth looking for nuance – and the counterpoint. Is Gen AI strictly a major productivity enhancement? Here are three things to consider:
The Velocity and Continuity of Change Will Be Historic
And that impacts productivity. The uptake of various gen AI tools is already far faster than any previous technology, and that velocity will be more disruptive to business and labor than previous technology leaps. Also, many new technologies, after being adopted, maintain slower evolutionary changes – invention, followed by incremental changes. Gen AI isn’t just disruptive day one, but will continue to evolve very rapidly, expanding and improving its reach and capabilities. It’s more about acceleration than velocity.
Previous examples include factory automation, PCs, the Internet, cloud computing. All were hugely disruptive, but factory automation takes time – decades – for full adoption. PCs took years for acquisition and “full” adoption. The Internet needed content and better and faster connectivity, and took years. Cloud computing, while lowering the barrier to entry dramatically, had to displace/replace huge investments in skills and capex, taking a decade to get to its penetration today – which is probably still less than a third of all data center computing. All of these changes gave enough time for reskilling, labor shifts, education. They proceeded at a velocity that human society could absorb. The overall unemployment rate remained relatively steady.
There’s good reason to believe this won’t be true for gen AI. There is very little barrier to entry – you can use it now, in many content-creation roles. Individuals can use it easily, and don’t need the enterprise to enable adoption. The benefits (and today’s limits) are readily apparent, and it will be uncompetitive for a business or knowledge worker to ignore, or delay.
Knowledge workers will need to rapidly reskill, and in many cases, change their roles. Where a knowledge worker may have been tasked to create content before, now they might be reviewing content creating by gen AI. Essentially a lesser skill, and essentially working for AI vs. the other way around. There will be knowledge work displacement both upward and downward across industries as roles are found for gen AI. Unlike previous technology disruptions, this will be fast, and will affect a lot of people, not always positively.
The continuation of change is also a factor. Unlike previous disruptions, this one will continue to advance. Today you may need to review content created by gen AI before publishing. Soon that will be less and less necessary. The breadth of content will continue to grow. Unlike previous disruptions, it will be difficult for knowledge workers to reskill and maintain space from growing gen AI capabilities, at the rate gen AI changes. The time it takes to reskill, relearn, job shift? All are continuous productivity reducers.
Supply and Demand Matters
Yes, content productivity of an individual or an enterprise will increase dramatically, but what happens to all that content, and what is the resulting value of growing content? For example, just how many news articles can an individual read each day? How many songs does the world need? Certainly, customizing content to niches and even individuals can increase content demand – but that’s still limited by the number of people. It seems likely that some industries will simply consolidate to fewer producers of higher-volume content – just how many song-writers can be economically viable if an individual can produce several new songs per day? Or, if gen AI can produce without any/much human involvement?
So an individual knowledge worker can be much more productive, but that likely means fewer knowledge workers – at least doing that kind of creation. Maybe far fewer.
What about the productivity cost of consuming more content? Again, we have the example of smartphones, social media and the Internet. Is the productivity of the consumer – the amount of useful information consumed per hour – better? Helpful? Does it drive more economic activity? It can, but the opposite is also true. We spend more time consuming a higher-volume of low-value information than we did ten, twenty years ago.
One way to handle the increased volume of content coming at you is to use tools to curate and filter it. Gen AI will have a role here, too, to take the firehose of content, use your preferences and filters and patterns, and present a subset for you to consume. What just happened? We used gen AI to ramp up productivity on the production side, and then we used gen AI to whittle it right back down. Creation productivity does not equal consumption, and gen AI might simply negate itself in some content flows.
Gen AI will become a powerful marketing tool, evaluating the patterns that are successful for manipulating individual opinions, acquisitions, etc. Humans simply can’t simply evolve at the speed of technology influence. Confirmation bias is an evolutionary weakness when technology is used that makes it easy to manipulate people. There will be much more effective influence wars for individual spending, votes, decisions, actions, etc. There will be no source of “truth”, but a growing multi-polarization, tribal beliefs, and a consumerization of truth – along with intense competition for influence, powered by Gen AI. Influence wars will grow, not shrink, and this friction is wasted productivity, where people and corporations will pull in opposite directions, and where investments both in production and consumption of influence will light fires and put them out, canceling out each other’s “increased productivity”, reducing net productivity. This will be as inoffensive as competing and sharper product marketing, and as threatening as deepfakes that inspire violence – even societal upheaval.
The net? Gen AI has tremendous potential to increase an individual’s productivity, but wide use of gen AI will also have productivity dampening effects – due to ongoing labor upheavals, gen AI use for both supply and demand “productivity”, demand limitations, and more friction and wasted energy in society. Gen AI isn’t “bad”, but “good” isn’t automatic. We will need to work hard to channel its productivity benefits, mitigate the waste, manage/regulate wasteful or dangerous content, invest in the ever-changing role of humans, and ensure that individuals don’t become casualties.
Again, I’m on-board – but with eyes open.