A POV: Generative AI & Commerce

By Heidi Bailey – Integer’s Group VP, Futures & A.I.

AI is everywhere! Specifically, Generative AI (Gen AI) is everywhere and already making a significant impact in every single industry, from retail to health to marketing and beyond. Entire industries and new companies are forming around this technology at such an accelerated pace that by the time this paper is posted, it will already be old news.  

It may seem that this is a new technology, as most people were introduced to Gen AI when OpenAI released ChatGPT back in November 2022. However, there have been many people in the AI industry working for many years to advance Large Language Models (LLMs)—refining, defining and making them more efficient for general marketing consumption. 

But what exactly are LLMs? They are a type of Foundational Model that focus on language.  Foundational Models, in turn, are neural networks that read and train on massive data sets so they can see patterns and learn and formulate rules without being told what those rules are. The output of these rules has been launched into usable applications like ChatGPT and Google’s Bard, among many others.  

This is what makes them so powerful. For the first time, models can generate novel content and keep track of context, meaning and its relevance to the market—and it can do it in a way that is incredibly useful for people across all industries. At this moment, and as far as we know, it doesn’t really understand what it is saying. It is merely predicting what to say next based on what it has learned from the training data. That is why these models sometimes “lie” or “hallucinate” as they draw the wrong conclusions or repeat untruths based on the content they were trained on. 

Keeping this in mind, and, of course, this is changing hour by hour, Integer has looked at both the positive and negative impacts these models can have in driving growth. 

Positive Impact: 

  • Potential positive impact of Generative AI in the Retail Industry $9.2 trillion expected by 2029 per IHL Group. 
  • Models can be trained, with guardrails implemented, and limited to specific commerce use cases to mitigate risks. 
  • Enhance the shopping experience both in physical and digital spaces by offering more personalized experiences. 
  • Achieve greater efficiencies in content creation. 
  • Conduct an in-depth conversation with site visitors. 
  • Deeper emotional and contextual media targeting capabilities. 
  • Integrate into current websites and apps to drive more engagement. 
  • Generate multiple variations of your current advertising copy and/or product descriptions.  
  • Translate website content quickly, offering the ability to deliver content in multiple different languages
  • Enhance search as it allows for complex requests returning detailed results.  
  • Faster software development. 
  • New product generation and development. 
  • Compile and summarize content from multiple websites into one search output, making it easier to do online research. 

Negative Impact: 

  • Models can answer questions and summarize thousands of websites into a single chat that people might typically find on websites, which could lead to decreased traffic on current sites. See future of search according to Google
  • Human displacement is a real concern, with a recent study by Goldman Sachs predicting job losses upwards of 300M people globally in the next few years.  
  • Biases remain in the training data sets of most of these models, so using them to create novel content at this time should be done with extreme caution and always with human involvement. 
  • Current legal battles around the copyrighted images and text that were used to train the models.  
  • Data Privacy, if the models are not being used correctly. 
  • Deepfakes which could take the form of influencers without their knowledge or approval. 
  • Cybersecurity problems, where any code inserted into these models could be exposed and fall into the wrong hands. 

At Integer, we have been working on delivering various commerce AI tools to our clients for several years and are actively exploring and experimenting with the many Gen AI tools in the market today. Additionally, we are building proprietary tools so we can understand and continue to bring forth the positive and negative impacts these models have so we can mitigate risks and move forward with driving growth for our clients.