According to research by PwC, artificial intelligence (AI) has the potential to contribute $15.7 trillion to the global economy, and boost GDP in local economies by 26% by 2030.

Previously, we highlighted 9 key skills for AI engineers. In this article, we’ll highlight how these skills can benefit your organisation and share practical use cases for production-ready AI systems. 

Use cases for machine learning (ML)

ML examines data for patterns. Specially trained models perform complex tasks, such as categorising information, image recognition, or making predictions based on past datasets. And because the technology is capable of learning, the results only improve and become more accurate over time.

In the media, we've witnessed high-profile use cases for ML tools, such as the AI software that can diagnose breast cancer 30x faster than a human doctor with 99% accuracy. As customers, we encounter website chatbots that offer AI-powered assistance to answer questions and direct us towards useful information. And as consumers, we experience the impact of ML tools used by financial institutions to flag potentially fraudulent activity on our accounts.  

In business, there are several applications for ML tools, which will help to smooth working processes, improve efficiencies, and boost productivity. For example:

Translation is a natural fit for ML tools. Today, global marketers need to invest in translation services from specialist agencies to communicate with their audiences across different countries and cultures. An ML tool is capable of performing the same task quicker, which relinquishes budget for additional campaigns that boost engagement.

Within product development, ML tools can also help you to produce visuals without needing to build a prototype. This can be particularly useful for user forums. In a few simple prompts, users can produce a quick mock-up of what a new feature could look like or to show a product in different colours/styles to gather feedback, which aids decision making. 

One use case on a fast growth trajectory is speech recognition. Able to integrate with more electronic devices and enable voice search, the market continues to grow at a CAGR of 14.6%. For working professionals 'on-the-go', it's an essential feature to be able to stay productive when away from the office. Furthermore, it aligns to AI personal assistants - a market that is growing at a CAGR of 17.3% and projected to hit £242bn by 2023. 

Customer churn is one area with the potential to significantly impact organisations. Every sales team knows it's easier (and cheaper!) to sell to an existing customer than acquire a new one. ML tools can help by identifying customers who are likely to leave, giving you an opportunity to implement a 'white glove' service to create stickier relationships.

Furthermore, ML tools can assist with more accurate forecasting. By analysing your historical datasets, and combining those insights with external market information, you can see when your business is likely to encounter peaks/troughs in demand, and plan accordingly. 

If your organisation owns a lot of machinery or capital assets, ML tools can help you stay on top of maintenance by predicting when equipment is likely to fail, or when new parts are required. Keeping systems up and running is essential for your business to avoid unplanned downtime.

Product management for AI

According to Gartner, traditional approaches and skills are inadequate when it comes to deploying and managing production-ready AI systems. While McKinsey claims that for organisations to succeed in the AI era, engineers need to extend their skills beyond technical competencies, to master broader “upstream skills”.

AI product management requires you to balance innovation and practicality to ensure the output is useful, usable, and used. 

During product development AI can cause the product to underperform, For example, poor quality data, which is missing, incomplete, or irrelevant, can impact the outputs your AI system will generate. An infamous example is Microsoft's AI chatbot 'Tay', which expressed offensive remarks on X (formerly Twitter) due to the poor data quality it learned from. This led Microsoft to pull the product just 16-hours post-launch.

Additionally, AI product managers must consider ethics to account for and challenge potential bias in data. This can be conscious, as in the example of iTutor Group Inc., which purposefully programmed its AI-powered recruitment system to exclude female applicants over 55 and male applicants over 60, regardless of their qualifications or experience. Or it can be subconscious, as in the case of Amazon, where historical data sets showed a male dominance at the company, which was perpetuated by its recruitment tool when it penalised female candidates in preference of male applicants. 

AI product managers also need to have effective governance in place to retain control of the technology. Generative AI (GenAI) tools are known to have accuracy issues. By its own admission, ChatGPT says it has issues with a lack of real-time knowledge, overconfidence and plausibility, as well as a limited understanding of nuance, specialised domains, and ethical and safety concerns. It even advises users to work with caution and cross-reference answers - something Levidow, Levidow & Oberman failed to do. When two of its lawyers used the tool to aid their research, ChatGPT invented fictitious cases, landing the firm with a $5,000 fine.  

With appropriate governance in place, these horror stories would have been easily avoided. AI works best when it supports humans in their roles, which means a person must always retain oversight of what the tool is doing, as well as check the outputs it creates. When using AI tools for decision making, AI product managers need to be able to explain how they reached a certain conclusion. It's only when we have transparency that we can have trust in AI products, because we can trace the reasoning behind a decision and the specific data points used.

Compelling use cases for GenAI

In the UK, nearly a third of the population (29%) uses GenAI. Compare that to the US where nearly half (45%) the population, and India where nearly three-quarters (73%) of the population are using GenAI, and we're falling behind the adoption curve.

McKinsey estimates that 75% of the value GenAI use cases could deliver falls across four areas: customer operations, marketing and sales, software engineering, and R&D.

From boosting agent productivity to enabling self-service, GenAI will significantly impact customer operations. For example, Salesforce developed 'Einstein AI' to enhance its customer service. Able to understand user intent, the tool generates personalised replies by pulling in information from the company's web links, knowledgebase and CRM. Even though the reply is manually checked by an agent before sending it to the customer, it significantly reduces response times and frees the agent to work on other tasks. McKinsey estimates that applying GenAI to customer operations could increase agent productivity by 30%-45%.

Personalisation at scale is a big driving force for sales and marketing teams. At L'Oreal, the company has embraced GenAI to offer its customers an AI-powered beauty advisor that performs an individual skin diagnosis before sharing tailored product recommendations. GenAI is particularly powerful when it comes to product discovery and search personalisation because it uses Natural Language Processing (NLP) to understand user intent. This means the customer isn't forced to think about how the underlying database is structured when crafting their prompt, and you can prioritise results based on inventory levels or profitability. McKinsey estimates that applying GenAI to sales and marketing could increase productivity by 5%-15%.

Additionally, the direct impact of Gen AI on the productivity of software engineering could range from 20%-45%. This is due to reducing the time spent on coding drafts, correction and refactoring, as well as root-cause analysis and generating new system designs. For example in a recent study, software developers using GitHub Copilot complete tasks 56% faster than their peers. 

Finally, within R&D, the use of GenAI within design is estimated to save between 10%-15% of total costs. In the pharmaceutical industry, GenAI is poised to revolutionise healthcare. Typically, drug development takes about 10 years. However, one biotech company called Insilico Medicine, used GenAI to create a new drug to treat a rare lung disease. It took 30 months and cost just 10% of the usual investment to develop. 

The best way to acquire AI skills for your organisation

With Cambridge Spark's AI Engineer apprenticeship (L6) you now have a new path to AI expertise - one that delivers real-world skills to create the next generation of talent. It can be funded via the apprenticeship levy and because it’s assessed against the Level 6 Machine Learning Engineer Standard, your team can acquire specialist, recognised AI qualifications. To date, our apprenticeships have generated £350m+ in ROI for our clients. Discover what Cambridge Spark will do for you.