AI Talk: How GenAI can enhance your organisation

On 26 November 2024, AI Talk host Kevin Craine was joined by Jamie Grabert, President/Co-Founder, The Consultancy Group; Chris Surdak, Managing Director, Advisory,Frontier Foundry; and Brenna Sniderman, Managing Director, Center for Integrated Research, Deloitte.

Views on news 

Forbes spoke with cutting-edge practitioners from IBM, The Estée Lauder Companies, Inc., P&G, Smart Design, and GenAI to learn about their GenAI and AI deployments. Use cases include trendspotter tools, hair dye development, the use of sentiment analysis in product development, optimising events for sustainability and personalisation at scale.  

 

The article suggests that all companies should carry out a thorough review of relevant, proprietary, obtainable, internal data, and potential external data that can significantly enhance the quality of the firm’s GenAI output, and assess whether the company should create its own RAG (retrieval-augmented generation) AI model, which can provide capabilities that are more customized for that firm and can better protect data privacy.

 

Marketers can use GenAI to better tailor their messages to user needs and they can also have a better view of how their messages are landing.  GenAI can amplify the capabilities of humans as storytellers and tool users but it must be closely monitored whether the business gets the results from a use case that it originally aimed for. 

 

The challenges of GenAI deployment

Businesses must give access to their employees to GenAI tools to prevent them from using ones that aren’t sanctioned by the company. GenAI uptake is the highest when it’s embedded in tools that the workforce is already using. There is also an element of trust, which can be strengthened by employers demonstrating to employees how GenAI tools can help them perform their tasks.  

 

GenAI plays a huge role in knowledge management but it’s key that a human is always in the loop. Use cases here range from large scale content generation to see what sticks to research assistant. As you get different answers to prompts formulated differently, compliance issues may arise when it comes to reproducing results in a predictable way later. To ensure that GenAI and corporate data is used properly, a business must perform audits regularly. To avoid anxiety about losing jobs to AI, employees should be reassured that they can stay in their position but they must go through some upskilling too to learn how to use new AI-powered tools. 
Compliance with regulations is a challenge when it comes to AI deployments.

 

Regulations vary regionally and tend to lag behind technological advancement. Organisations also struggle with protecting the data that is fed into these models from a cyber security perspective. Because of these challenges, businesses still see data as a barrier rather than an asset. AI is inherently ungovernable, which doesn’t mean we shouldn’t try and improve its governance. Also, if there is no human that can tell a GenAI generated answer is wrong, the model will reinforce the wrong answer. A good strategy is to use LLMs as source code and build a smaller language model on them. You don’t need your enterprise model to be trained on blogs and social media feeds to do the job for you you’re your proprietary data. A company’s AI governance framework should include enhanced data security, data quality practices, rules about the frequency of updating your data, a  policy governing cloud service use and integration with other tools. It must also touch upon talent and how the business ensures that its employees can work with AI.

 

There should also be a policy on whether the business is ready to use intellectual property content for training its model. The risk that companies must take is that they deploy t the governance of which may change in a few years’ time, the direction of which is still impossible to predict. It’s also part of governance to establish how the business deals with hallucinations and what process sgould be followed when the AI tool goes off the rails. 

The panel’s advice

• As the workforce familiarises itself with GenAI tools, new use cases will emerge. 
• Use case selection is critical to getting value out of GenAI tools.
• As a leader, you must explain how an AI-driven future looks for your company, as well as what role your employees will play in it.
• The seven Rs of AI governance: real, reasonable, regulated, reliable, resilient, reversible, recordable?

For Deloitte’s quarterly reports on GenAI, click here.