Original article was published by Alexandre Gonfalonieri on Artificial Intelligence on Medium
The need to automate retail stores has become a priority due to the recent sanitary crisis. Indeed, it has become essential for retailers to adapt their business models if they want to survive. In this context, I worked on a project related to AI decentralized autonomous organizations (AI DAO) and retail stores.
Can a store become independent? Can you invest in a store that does not require human workers? Can we make a store more intelligent by using distributed artificial intelligence?
In this article, I will explain why existing retailers are at risk and must urgently invest in new business models, explain the key role of AI DAOs and how retailers could benefit from them.
Retail stores as we know them will disappear
Due to a series of lockdowns and the rising threaths from online competitors, retailers are left with no choice but to embrace automation. In general, the way we shop is changing.
To put it in simpler terms, the future of retail is about automation, smaller stores, phygital experiences, and more flexible workforce.
Some retailers have decided to transform their stores into storerooms and delivery depots (1) for businesses that have moved entirely online. Others have started working on semi-autonomous stores.
From what I have seen, the automation path often requires several computer vision algorithms. Leveraging computer vision helps retailers collect enough data to better understand their customers.
The main strategic point is to build the best customer experience possible. The only way to achieve this is to gather as much data as possible about customers and how they interact with all products.
AI DAO For Autonomous Stores
This automation can go a step further: AI decentralized autonomous organizations.
A DAO is a decentralized model for businesses. It is made of smart contracts governing how the business should operate. All of a “DAO’s business decisions and financial actions are recorded on a public, immutable blockchain” (2).
Each participant/investor in a DAO gets a vote on how the company is run. This blockchain is accessible to all investors/token holders, making it easy to share all the information with every investor.
An AI DAO would be a DAO using multiple AI agents (Swarm intelligence). Other ways exist to create an AI DAO (3) but we have decided to focus on this one .
According to S. Ponomarev, A. E. Voronkov, “Multi-agents systems communication is a technology, which provides a way for multiple interacting intelligent agents to communicate with each other and with environment. Multiple-agent systems are used to solve problems that are difficult for solving by individual agent.Multiple-agent communication technologies can be used for management and organization of computing fog and act as a global, distributed operating system” (4).
A multi-agent system is simply a form of M2M (Machine to machine) communication. In other words, it is a loose ecosystem of various communicating AIs.
This form of M2M communication allows learning AIs to actually exchange information and solve complex problems faster. In the future, autonomous stores will interact with other organizations (providers, etc.) using AI agents in a global M2M ecosystem.
Machine to machine (M2M) is direct communication between devices using any communications channel, including wired and wireless.
This need for AI DAOs also comes from the fact that AIs exist only in silos. Current solutions are often only capable of operating within a specific company, infrastructure, or department. This lack of interoperability severely limits the development of fully autonomous stores.
With coordinated AI agents, retailers can outsource a significant amount of network decisions, including specific issues with governance, voting, security, etc.
Imagine a store run by an AI decentralized autonomous organization. Contractors around the world could be responsible for the maintenance/logistics of autonomous stores owned by the AI DAO and could be paid directly by the AI DAO itself.
This same AI DAO store could use money to autonomously restock the products, hire the services it needs (cleaning, security, logistics, …) and pay on its own.
Furthermore, as more people invest money in this store, all the users will have a saying and a vote that will influence the decisions taken by the AI DAO.
Such a store would leverage several AI subfields (NLP, Computer vision, …) to adapt to customers needs, tracks spending and preferences. I believe that this entity would be based on the evolutionary theory. AI DAOs will never be seen as “finished”. The services, products and prices will always evolve.
Beyond the automation of most tasks, the decision process will be quite unique.These decisions can be made using smart contracts.
A smart contract is a piece of computing programming that executes automatically when certain conditions are met (5).
According to Joe Liebkind, an AI DAO “with smart contracts can automate tracking of inventory, creating and payment of bills, incoming shipments’ scanning, etc” (6).
By adding AI, DAOs will create new business models. Soon, I expect to see a growing number of fully independent businesses in which everyone can invest. Perhaps retailers will stop considering their customers as only consumers but also enable them to own a piece of an autonomous store.
We might enter a new era in which the average customer can also easily become an investor. This new business model will apply to many other organizations. This may include decentralized hedge funds, decentralized public utility providers, etc.
AUTOMATING — Yes, but what and how?
AI DAOs must be based on several AI subfields to cover as many areas as possible. Here are some areas generally that can already be covered by AI :
AI Decentralized Autonomous Organizations Business Models
By leveraging Generative Adversarial Networks (GANs) and 3D printing, we can imagine a store that analyzes market trends in real-time and decides to manufacture the most trendy item on the market. The autonomous store would collect sales data and determine if a given product is a fit based on the average customer that visits the store.
This same autonomous store could also leverage Machine Learning-based algorithms for dynamic pricing.
I predict that GANs will be used a lot in combination with AI DAOs. Indeed, several industries could benefit from this combination. Autonomous marketplaces could be created to sell items made by GANs.
Recently, a web platform (7) managed to make a Generative Adversarial Network (GAN) generates art. This painting was later sold the revenues were used to pay for the resources used.
A generative adversarial network (GAN) is a class of machine learning frameworks designed by Ian Goodfellow and his colleagues in 2014. (8)
I also predict that some autonomous stores will only be made of interactive screens. We can imagine interactive walls that dynamically change their content as customers walk by or based on what’s happening online. This will enable new brands to have access to “physical” stores. A random totem screen located in the street would become an independent store.
Based on my experience, plenty of questions remain around the effectiveness/safety of existing technologies and whether they can scale upwards from small stores to large ones.
One of the main challenges is to see precisely how many times human shop assistants are needed to resolve issues. Moreover, the customer privacy is also something to take into consideration. What data can be collected? How do we adapt our data strategy in several countries?
Another issue is the lack of data. Indeed, to train AI and develop machine learning algorithm requires sourcing large scale of data which is extremely costly. Beyond data augmentation, one way of resolving this issue is to use synthetically generated data.
Synthetic data is information that’s artificially manufactured rather than generated by real-world events.
I believe the development of autonomous store will follow the same path as autonomous vehicles. Today’s challenge is about converting existing stores into semi-autonomous stores.
Stakeholders need to find practical solutions to make that transition and reduce the cost of conversion. In a few years, AI DAOs could be a solution.
Data importance for retail stores
Most major retailers have now already started some proof of concepts to better understand their customers using computer vision. The main challenge for them is to better understand new shopping habits while reducing all the costs associated with a point of sale.
Obviously, with the advent of online shopping, the risk is to end up with many irrelevant physical point of sales. How do we create new revenue models and customer experiences? How do we combine the best practices of an online and offline store for enhancing customer satisfaction.
AI DAOs could help create new relationships with your customers. From simple customers to investors in your “own” autonomous stores…