Everywhere you turn, Gen AI is stealing the spotlight. From thought leadership articles to industry conferences, this buzzword is on everyone’s lips. But amidst all the chatter, it’s easy to get lost: What exactly is Generative Artificial Intelligence? Why is it being hailed as the next big ‘game changer’ for businesses? And how do you take advantage of it in your organisation?
We asked three of our specialist digital & innovation agency leaders for their views. They’ve shared their actionable insights to help you bring clarity and uncover how to harness the power of Gen AI to make a positive impact on your business.
Kath Blackham, is a conversational AI specialist and founder of Versa Agency & Versa Connects. Simon Tobias from Tobias, focuses on Human Centred Design, Innovation and Capability Building. And Palo-IT’s Chief Strategy & Impact Officer, Xavier Rizos, brings a corporate innovation focus to this topic.
What business leaders in Australia need to know about Gen AI according to these Digital & Innovation Agency Leaders:
1 - Gen AI won’t replace your people, but it can supercharge them
There’s a long-standing fear that AI threatens our jobs and will replace the need for people in our businesses, and the rise of Gen AI has heightened that fear for many. According to our specialists, this is a misnomer (at least for this latest wave of AI). In fact, there is a real opportunity to leverage Gen AI to help boost the capacity and output of your team.
“Deploying Gen AI tools across your team is like giving every team member their own intern,” explains Kath. “Imagine if you could give every single person in your team a Gen AI assistant and let the tool do the repetitive or basic tasks that suck employees’ happiness and slow them down.”
“Microsoft’s Copilot, makes developers, for example, between 50 and 67% more productive. Notably, it also has an employee engagement and retention impact - developers report a higher level of happiness in their jobs because they had more time to focus and think.”
Copilot isn’t just for developers. If your organisation uses Microsoft, keep an eye out for the upcoming release of Copilot 365 - a digital personal assistant for the Microsoft Office suite that can generate actionable tasks for you during a Teams meeting, or create images in your PowerPoint slides to help bring to life the story you want to tell. If you work in a large organisation, you may already have access to it as part of an early user trial. And if you are not a Microsoft user, Google has a similar offering, Duet AI, available already in its Google Workspace suite.
Simon echoes this view about Gen AI presenting an exciting opportunity to empower your team. “I think of it as an enabler for customer-obsessed organisations. It’s a way to augment ourselves - whether that’s in the pursuit of better understanding our customers or in responding to the insights we uncover.”
2 - Identify where and how Gen AI could benefit your business most
While it’s true that Gen AI can help boost your own productivity and that of your team’s, what we’re more interested in covering with this article is the potential impact and opportunities associated with Gen AI for your business model (think larger scale innovation and business transformation).
To drive tangible outcomes for your organisation, it’s all about finding the right use cases for Gen AI. You need to understand how you can use Gen AI to better solve your customer problems, Simon points out. “Smart adopters will be using the new generation of AI primarily to solve old problems. They’ll be supercharging their existing products and services this way.”
Simon further explains that “enterprises still struggle to respond to the increasing volume of customer data and insight they’re gathering. There’s a lot of value left on the table.”
The potential applications in an enterprise are vast, from customer service and operations, through to content and product development, marketing, and risk management. To identify the most suitable opportunities within these functional areas, Xavier suggests that business leaders “focus on very well crafted customer insight studies, well defined problems to solve and opportunity spaces using good design thinking methods combined with strategic intent and what matters for your business.”
For Kath, “apart from the hugely exciting advancements in productivity and how we work, for me the most obvious use case by a country mile is CX or more specifically the call centre.”
Kath and her team have worked to deploy the technology into call centre environments with clients across various sectors including retail, BPO, finance, insurance and superannuation - improving their customer experiences while also reducing operating costs.
3 - Start small, experiment often
“Find somewhere to start and play with Gen AI on your own,” says Kath. “Then find some compelling use cases for your enterprise - look for Proof of Value, don’t just build out a Proof of Concept. But either way, keep it small at first.”
Simon echoes this point: “You need to run targeted experiments but these need to be focused on solving a customer problem. I also think that it’s useful to break these experiments down thematically into categories, for example: new products and services; enablers for better customer experiences; or tools to augment creative output or democratise innovation.”
A great example of such a project that Tobias has been working on for a global Financial Services brand involves leveraging customer and employee insight to define a future state vision, target operating model and roadmap for achieving their North Star. One of the critical enablers of this transformation project is a series of AI experiments to support the future state CX.
Xavier points out that it’s less relevant today to be questioning whether Gen AI is going to lead the most superlative disruptive scenarios, or whether it will under-deliver on investors’ expectations as seen in previous Tech bubbles. The nature of this technology breakthrough and its adoption across sectors already tells us that a reasonable middle-scenario with efficiency gains is already happening. This reinforces the imperative for organisations to start experimenting sooner rather than later in order to make sense of it.
“As the rest of your industry comes up with lower cost to serve underpinned by Gen AI, missing out of those productivity improvements puts the business model at risk. The strategic two-step way of thinking about it is: protect and defend how your organisation brings value today; then do not be outmaneuvered in the future as Gen AI gets adopted and new business models emerge. To do this, we believe running considered Proof of Value and Proof of Concept experiments, as opposed to large projects, is the way to progress,” cautions Xavier.
4 - Put the right guardrails in place for real success
Of course, data security and risk mitigation needs to be top priority for any enterprise experimenting with Gen AI. It is vital to ensure you are working in a secure environment.
“Before you do anything,” advises Kath, “put boundaries around using LLMs like ChatGPT. This is where it may be helpful to consider employing tools like Katonic - a machine learning operations platform that helps you solve the problem of data security & accuracy of results by fine-tuning and installing Gen AI models on the organisations.”
Ask questions like:
- Which Gen AI tools are safe/best for your enterprise?
- Does your organisation only want to use a closed version of an LLM, like Microsoft has created, for example, or are you comfortable with using open data sources available to anyone?
- How can you harness this technology without compromising your enterprise’s IP?
The risks aren’t only in relation to data security and privacy… There are also potential issues around data accuracy. Simon explains, “generative models are still just predicting a plausible answer to a question you’ve posed through training. It’s not always accurate. It can hallucinate - making up things that seem plausible but are only 80-90% correct. This is generally acceptable in writing, because in conversation, we usually accept a lot less accuracy than this.”
So, it’s not great to unleash ChatGPT on anything that requires precision like legal text unless it’s drawing on a ‘walled garden’ source of authoritative data - also known as a closed LLM like the one Kath describes above.
Xavier adds that “Gen AI is not going to replace rules-based algorithms which use maths and logic. They are all different types of computing methods that accomplish different things. We are not going to throw out one in pursuit of the other. They will ultimately all work together and integrate on specific use cases: for instance leveraging natural language processing when a conversation interaction is needed, and still using hard logic when computing financials is necessary. That’s how exponential value will be unlocked.”
ChatGPT gets all the press, but it’s not the only LLM in town. The paid version has a setting that can be used, where it will ‘forget’ data you use by turning off chat history. Or you can use something like Azure AI with your own secure cloud instance, trained only on your data and locked down from the outside world. Or create an instance of the OpenSource LLaMA 2. Or create your own model from scratch…
It’s not only privacy, consent and IP we’re talking about here: consider vulnerable populations and the effect of data leaks in this context.
As with all large scale tech revolutions, data protection and privacy is only one side of the coin. The other is how we design and build our systems ethically.
Simon makes a great point, when he reminds us that “bias gets amplified by technology that scales. These systems are reflective of humans. You need to be explicit about making sure the models you’re using are reflective of the values you want from your company. An important consideration is to concentrate on automating transactional things, but keep humans involved in decision making that will impact others.”
While there are fundamental steps you need to take to ensure that you’re working in a secure environment, Gen AI can also help you remain compliant and consistent. An example of this is when Versa Agency worked with a major pharma client to help streamline and standardise responses offered off the back of customers’ FAQs and support queries. It is ideal to use a closed LLM in a highly regulated industry, like Financial Services, because it delivers consistent responses and outputs, which are also easily traceable - helping ensure your service to customers is compliant.
5 - Balance experimentation with a longer-term vision
When you are looking for opportunities to introduce Gen AI, Xavier reminds us that we should not “just identify the opportunities associated with Gen AI, also look for the risks it could pose to your business model and/or industry… If I were running a business unit, I would want to understand how the dissemination of Gen AI will threaten - and how I will have to protect - my product, my business model and my customers over the long run. Not only what it can do for them…”
In the short term, experimentation will act as an enabler. You will, however, need to develop a long-term view on Gen AI in your organisation and build out the relevant policies. Because it is such a big potential disruptor, your enterprise needs to take a stance on it.
“You need to grow your knowledge management capability,” reminds Kath, “so the systems and products you build using Gen AI have all the information needed to create amazing experiences for your team and you have to have people that know what they are doing when they build out the architecture.”
The longer term vision also needs to consider our human capability to adapt, says Simon. “It’s not skills, roles or jobs - it’s about creating the environment, and the mindsets, skills sets and toolsets to operate in new collaborative and inclusive ways of working.
Empathy, agility, creativity, adaptability, critical thinking, communication and ethical practices are all hallmarks of the future workforce - and you will need to start nurturing these within your enterprise.”