It is no secret that AI is currently at the forefront of industry-wide decision making and the retail sector is no exception.
The AI industry was worth $81.3bn in 2022 and the market is expected to achieve a compound annual growth rate (CAGR) of more than 21% between 2022 and 2030.
As well as its exponential growth, AI can offer retail businesses transformative technologies such as AI-assisted surveillance and digital shopping assistants.
However, harnessing these technologies is easier said than done. Many businesses are unsure about how to capitalise on AI and whether it poses any dangers along with its benefits.
To find out, Retail Insight Network spoke to AI platform Coveo’s senior director of e-commerce marketing Sergio Iacobucci about how AI can be capitalised on.
What will happen to retailers who do not utilise AI effectively?
Retailers who do not utilise AI in their operations will find it impossible to keep up with competitors leveraging the technology to improve efficiency, provide a better customer experience, reduce costs, and identify new opportunities.
The opportunity costs are very high: reports suggest that 15-35% of incremental revenue opportunities are available to brands adopting AI. It’s not far off the same number for increases in profits too.
What dangers does AI pose to retailers?
The risks associated with AI in retail are largely related to how it is used and implemented. Retailers and consumers need to be aware of these risks and take steps to mitigate them. This is done by investing in robust data privacy and security measures, ensuring that AI systems are trained on unbiased data, and using AI as a tool rather than a replacement for human judgment.
You need to ask questions of your vendors: what models do they use, what type of data trains the models, and what biases might exist. A/B tests are the best way to ensure you get the desired results.
The potential danger areas for businesses and AI, outside of not adopting it at all, are:
- Privacy and security: It’s great that AI systems can trawl through lots of data which is very insightful and useful for the business and the customer when they can do this with customer data. However, it naturally poses some risks if handled incorrectly with enterprise needs in mind.
- Data bias: AI uses the data you give to learn about that given dataset. If that dataset has biased data within, it will bias the output from the models. This is where mature AI practices come into place and working with the right internal teams and vendors becomes critical.
- Workplace change: Changing how tasks are completed changes the fundamentals of certain roles. People rarely like change, hence why AI is posed as a risk. Businesses will need to consider how to utilise their people for higher value tasks that a machine will struggle to do. This is largely where human creativity and innovation come into place.
These risks can be mitigated with education, human oversight and diligence. The biggest risk is not knowing about them at all.
What can be done for retailers to better understand how to utilise AI?
Scepticism around AI in businesses today exists. Coveo recently surveyed more than 75 enterprise search tech professionals in retail, and this was a common theme.
Not having the right internal expertise creates a lack of resources to effectively deploy and manage (47%) AI projects. They also say they need more training (47%) and want to retain control over search ranking and results (37%). Here’s how it can be done:
- Use all the resources from trusted establishments to educate themselves generally on AI
- Identify the high volume, low value tasks within the organisation that would be good candidates for testing AI implementation.s
- Build a small, agile team to look into these use cases and be accountable for delivering on these projects.s
- Select an AI vendor (within the given business function where you’ve identified the correct projects) to help you understand the problem better, communicate within the organisation and implement tests to learn how AI can free up people to do higher value tasks.