It was a sunny Saturday in Sydney and I was enjoying the Tomato Food Festival. Vibrant tomatoes of all sizes adorned the tables at the fair, with colours ranging from vivid green to burgundy and dark plum. An Australian farmer spoke passionately about tomatoes, ruefully claiming that there used to be 4,000 varieties but this had shrunk to 3,000 in the past decade. He claimed that the others had been gradually ‘bred out’ by the big supermarket chains. Together with other volunteers, he was there to educate consumers on the types of tomatoes we never knew existed.
The story was sadly familiar one – a niche producer argues that larger producers have limited consumer choice to suit their scalable production needs. As the growth of intelligent automation and artificial intelligence (AI) sweeps the world, it raises a further question – how will AI play into the development of scale, and indeed how will that impact on global business?
Scale as leverage
Scale is the notion that a single idea can be projected to solve many needs. Most ideas are scalable in human society. Our language relies on scale and homogeneity of definition to communicate – my definition of a shoe, for example, must be similar enough to yours and to our neighbours. In business, scale is leverage, allowing a homogenous unit to be produced in great quantities for a low marginal cost.
In financial circles, a mutual fund is an example of a scalable investment solution. A single investment strategy can largely meet the needs of many investors, allowing a single manager to oversee billions in client funds. Increasingly, however, scale refers to the idea that a single resource (be it a manager or an algorithm) can manage 100 strategies of $10m each, in the same way that currently a single manager would manage a large $1bn fund.
Fragile scale and AI
Building information systems with the scale to solve today’s problems but which can also adjust their size and scope to solve tomorrow’s problems, is difficult. Especially when it is not clear what tomorrow’s problems will be.
The idea of scalability promotes the idea of building focused systems that solve a specific problem really well. This focus often comes at the expense of flexibility that would allow the system to be disassembled and rebuilt easily. In fact, the act of scaling itself can introduce complexities – complexity is seemingly the uninvited bed-mate of scale.
AI offers the promise of customisation and virtual flexibility. The ability to ‘rewrite’ the automation algorithm to suit the problem. As a result, AI should introduce flexibility intelligently by essentially keeping the product or service parts of the process flexible and dynamic. The role of intelligent automation is to fill in the blanks. This idea reduces the fragility of simple automation. By making the system more dynamic, AI should offer both flexibility and scale.
The ‘benign monopolists’
There are many societal impacts of intelligent automation, and most seem to revolve around the notion of creating giants of industry that symbolise the type of domination that comes from unassailable scale. Almost by definition, this pushes prices to zero, where most competitors are ‘starved out’ of the market. Unlike ‘low-balling’, there is no need for the surviving dominant player to raise prices and exercise monopolistic behaviour because they are already profitable at the lower price point. The only defence is to attempt to create a ‘value’ proposition that is not based on scalability – as our intrepid tomato farmer is trying to do.
However, AI raises the bar on what the ‘unique’ value proposition is, taking actions that used to be innately human and automating them. Customer service and human communication itself are currently close to being automated, and therefore, becoming scalable commodities. A similar argument can be made for services like medical, financial and legal advice. While all of these have a value proposition beyond the scalable component, like the experience of seeing a person, most consumers will be happy to receive something that is worth 90% of what they are used to, if they only have to pay 10% of the original price.
The winner-takes it all
The winner-takes-all proposition comes from the need for quantity to maximise revenues for huge scale firms charging low prices. As scale increases, prices approach zero and the quantity of the good or service required to make up for the falling prices needs to rise as well.
This creates a kind of ‘benign monopolist’, a company that embraces scale and uses it to create and maintain barriers of entry by keeping their prices prohibitively low for new entrants. In other words, scale here can become a barrier to entry itself – consider how many search engines you can name.
In this industrial organisation, AI is a tool used to maintain scale. An interesting example is Google’s recent foray into hospitality and media marketing, both of which have seen extreme falls in prices. In asset management, proliferation and aggressive pricing by passive fund providers is creating a highly concentrated market in funds with extremely low levels of fees and high levels of automation.
Scale begets more scale
Scale requires clean freeways (or distribution channels) to travel and proliferate. Enter the highway creators, platforms like Amazon or Alibaba, that create incredibly accessible and transparent ‘freeways’ that companies can use to buy and sell goods globally. Other than widening market access, these companies enable and encourage firms to try and increase their scale.
While physical shipping and transportation is fraught with delays and occasional taxes, information knows no such boundaries. Information can be ‘shipped’ anywhere in the world without friction, allowing companies like Netflix to effectively compete with and replace local television and media companies globally from one central location in San Francisco.
Analysing the megatrends: Fintech, China and banking transparency
Future freeways may develop that allow the connection between hospitals, allowing a single AI like IBM’s Watson to become the first ‘global doctor’. A service that could theoretically diagnose and suggest medication for any human being in the world, and one that belongs to a single company. And, if Watson is the best intelligence we have at the time, it will be the only one required. Medical research, law, engineering, financial services and many other sectors could suffer the same fate – replaced by the ‘highest’ form of automation on offer in each of these fields.
Homogeneity is not, however, always your friend. Ecosystems that exhibit excessive levels of similarity are fragile from an infrastructure and network perspective. While we may marvel at the highest form of AI that is globally distributed to solve a specific problem category, it equally ‘leverages’ the faults of that system as well. Cyber-attacks and similar disruptions are a strong counter-argument to the benefits of connected, homogeneous systems globally.
Scale and AI: Leverage with customisation?
A world of scale-focused companies may seem like a dystopian future, and very much the antithesis of a cottage industry or a sharing economy.
For physical goods it seems that we are still some way from all getting our furniture, tomatoes and shirts from the same entity. However, for entertainment, social feedback, advice, diagnosis or any information-based service, it seems that scaled services aided by AI, both as the service itself (e.g. an automated doctor) and as a customisation agent are here to stay and are likely to create even greater behemoths of technology.
The combination of AI and scale will have an enormous impact on the business landscape, creating industry giants and rewriting the very core of some industries. Barring some technology collapse, that is not going to change. Existing and new businesses will need to embed intelligence and scale deep into their business models to compete effectively, and they may need think very ‘big’ or very specialised.
More from The Travelling Quant: Introducing the machine hulk