Back to the future: How Artificial Intelligence (AI) is shaping the next digital economy

Policy Horizons Canada’s The Next Digital Economy report (2019) described a plausible future economy in which the various things needed to produce goods and services could be located and owned more broadly rather than concentrated in an individual firm.

An example of this production model was described in a 2019 vignette. It describes someone looking for a custom pair of shoes in 2029. Within a digitally intermediated distributed economy, they begin their search by asking their artificial intelligence (AI) assistant to find shoe designers. Their AI searches the Internet for suitable designers and requests design and price proposals from the designers’ AI assistants, before selecting the best option and finalizing a contract. The designer is paid when they fulfill a blockchain-based contract by sending the design to a 3D print shop close to the consumer. The print shop is paid when a drone delivers the shoes.

Five years ago, we expected AI to be a key technology in facilitating the connections between humans and devices, as well as between devices.

Since 2019—when IBM’s Watson was the cutting edge of AI—there have been rapid and significant advances, most notably in generative AI. This includes AI like GPT or Bard, which are based on a large language model (LLM), as well as Stable Diffusion or Dall-E, which are based on a deep-learning text-to-image model.

The expanded capacity of today’s AI suggests that we should update the 2019 model of a digitally intermediated distributed economy in three important ways:

• The 2019 model assumed that a human designer would be needed. The prospective customer described the shoe they want to the designer; then, using creative and artistic skills, the designer produces designs for the customer to review. Today, image generation AI tools allow someone with no creative talent or artistic skills to generate and refine designs from simple text prompts.

• Generative AI can write original computer code. This means that it could fulfil another task that the 2019 model assigned to humans: turning the shoe design into code for a 3D printer to manufacture the shoe.

• The 2019 model incorporated smart blockchain-based contracts, which assumed the involvement of humans with legal and coding knowledge. However, it seems plausible that generative AI could produce contracts and terms of payment for the value chain.

The rise of generative AI could accelerate the emergence of an even more streamlined version of the economic architecture described in The Next Digital Economy.

Generative AI potentially decreases the demand for expertise from certain creative and knowledge workers. Consumers could have access to generative AI that is encoded with some of that expertise, giving them greater agency over the design and production of goods and services that meet their specific needs.

However, there is no assurance that this would lead to reduced costs for goods or services. The economic value generated by the creative and intellectual expertise involved in design, production, and contracting would simply shift from designers, coders, and financial service professionals to the firms providing the AI services. Prices could remain high if a few dominant firms monopolize AI assistant services.

Ultimately, incorporating generative AI into the economy could erode significant fields beyond those typically targeted for automation, without lowering market prices. A few tech companies or platforms, rather than consumers, would benefit from potential productivity gains in this future.