Project Intro

This was a class project from Fall 2021 in which a partner and I were tasked in creating a mobile application that used artificial intelligence (AI) and/or machine learning that improved an activity. In our instance we created Price & Compare, an app that uses aids users in finding the cheapest available options in a quick and effective manner.

The initial problem we discovered was that shopping users have limited budgets and time in their busy schedules, and that they required an effective solution for determining the nearest locations with the most affordable options.

"Price & Compare" is an advanced application that allows the user(s) the opportunity to scan a product’s barcode. The user can also use advanced machine learning to receive a resourceful list of the nearest deals of affordable products. Through machine learning, ‘Price & Compare’ has a comparison feature that gives a list of related products based on the user(s) learned behaviour patterns. This feature is based on the user’s favourited products, giving the user(s) a personalized experience.

How It Works

Our app uses machine learning through its “Related Products” and “Daily Suggestions” features. Price & Compare learns from collected data to improve in accuracy over time and delivers suggested and related products through the use of this data. The machine learning in this app uses unsupervised machine learning - this uses an algorithm trained on data and finds patterns and relationships to make future predictions of suggested and related products for the user. These suggestions are tailored to the users’ specific interests and needs and encourage them to purchase the most suitable and affordable products based on their data.

Price & Compare also uses image recognition. The image recognition in Price & Compare can be seen through the barcode and product scanner. This image recognition involves a neural network that processes the individual pixels found in the product’s image. Once a photo of the product is in focus, the app then analyzes the pixels and scours the internet to find the stock of nearby locations to create a list of recommendations for the user.

What I Learned

Part of the process for this project had us identify potential consequences from this product, via an Iceberg Canvas. From our findings, we found discovered a few issues.

Firstly, we found that our app has the potential to take away demand from smaller businesses due to larger businesses having the ability to price their products for less. This may be as a result to larger companies having lower production costs, or smaller businesses not having a digital presence.

Secondly, our app would rely on the user having WiFi or mobile data to operate the app. Without connection to a network, our app becomes significantly limited.

A third issue we discovered is the rate of adoption. I used to work at a grocery store that switched to distributing weekly flyers via their app instead of having printed copies. This change left customers unpleased, as they had no interest to install an application. Similarily, why would some users adopt this technology when they could stick to what they know?

Finally, data collection was an issue. How would user data be collected and used? Would we sell this to advertisers? Could users be identified in a data leak?

Limits and Possibilities

The image detection used in the product scanner function of our app is a piece of technology that is not yet as advanced in the application. This will most likely be a legitimate possibility of the future and could revolutionize the shopping experience and customer behaviour.

We also ran into the issue of how the user could accurately compare the price of weighted items, such as produce or meat. Either image recognition technology, or other new pieces of technology would need to be advanced to provide a much more accurate price comparison.

Another problem relating to produce is the ability for our A.I. to distinguish between regular and organic. This would require input on the user’s behalf to get the most accurate price.

In the context of our product, we believe that implementing the ability to purchase within the application and bypass the tills would be beneficial and in line with our user statement. This was a feature we initially prototyped, but ultimately we decided that focusing on an application that offers price comparisons and related products would be more appropriate for the scope of the project, A.I. and Machine Learning.

In terms of how e-commerce will play into brick and mortar shopping in the near future. Image detection technology will revolutionize the way customers operate when shopping simply by using their mobile devices to find specific products and access additional information about the products. If a customer finds that a product is out of stock at a store they currently need to use an in-store terminal to have another location ship it to their home. In the future this customer will be able to use their smartphone to find a lower price at another location and order it electronically for an in-store pickup. The future will provide lightspeed e-commerce and a revolution of digital retail.