Generative AI, two years on from the advent of ChatGPT – What to do next

This article has been published as part of our QUT Business Leaders' Forum series. BDO has been a Primary Sponsor of this series since 2013. Learn more here.

My ten-year-old daughter, Safeerah, cannot believe my usual keynote presentations on Generative AI take me 45 minutes. “You ask any question you like, and it can answer for you - that’s it, Daddy, presentation done”, she tells me, shaking her head.

Her uncluttered young mind has led to this simple but elegant description of the core functionality and use of Generative AI applications such as ChatGPT after playing with it only a few times. Interestingly, she has already discovered a few limitations - it’s not always right or useful, she can be quite lazy with it, ChatGPT is much better than Snapchat’s AI, and some of the pictures Dall-e generates are not how kids would imagine them. But she has simply taken these factors in her stride and adapted her use of Generative AI accordingly.

As leaders running organisations, we, of course, must be more cautious in our use and adoption of Generative AI, but perhaps there are some lessons we can take from the way children like Safeerah adopt new things. Their curiosity and willingness to explore and ask questions while learning are traits we can all emulate.

One of the kindest things OpenAI has done for society is release an imperfect Generative AI product (ChatGPT) which they are continually developing. This has given us all an opportunity to ‘grow with the product’ and go along on the journey. It is important for us to take this opportunity and create a safe space for our team members to explore and learn how it can be applied to their work.

There are some risks in relation to Generative AI which must be considered. These include the validation and accuracy of answers generated, privacy and confidentiality of the information, and a wide range of ethical considerations we need to be aware of (e.g. bias). However, all these risks can be managed and should not be a deterrent to the adoption of Generative AI, at least at an exploratory level. Generative AI’s barrier to entry is much lower than previous technologies - you can simply ‘talk to it’. It is also unrestrained in the sense that each one of us can explore and use it in whatever way we wish to.

Ironically, it is perhaps this limitless application that is the biggest barrier to its adoption in organisations. We are so used to looking for specific examples and benefits of new products that we almost don’t know where to start with a product like Generative AI that can be applied across ‘everything’.

Research from MIT suggests Generative AI can save about 40 per cent of time on tasks and improve the quality of the output by 20 per cent. The race seems to be on to find specific use cases for Generative AI, and we will see a significant increase in the number of applications and products in the coming months. Some of these products will be visible while others will simply be embedded features in applications we already use. There are two big use cases of Generative AI at the forefront of discussions:

1. Creating a new document (e.g. a report or a proposal) based on the analysis of precedents and other relevant information within your organisation - the AI will trawl through your previous reports and relevant information, and draft a new document having regard to writing style, content and the context of the new requirements.
2. Guidance through a set of rules or policies within your organisation - e.g. ‘help me understand our company’s parental leave policy and help me apply for parental leave’ - the AI will trawl through all your relevant policies, guide you through them and answer any questions you may have.

There is also a third more general use case – Generative AI can be used by anyone in an organisation to help improve their work activities. They can brainstorm new ideas, create and analyse lists and explore the pros and cons of a certain topic. Unfortunately, the broad, non-specific nature of this use makes it more difficult to quantify the benefits than the two specific examples set out above. While the benefits of each individual usage example may not be as high, they will be substantial collectively.

As a society, we are only in the early stages of discovering Generative AI and its future potential is significant. While there is a lot more to come, it is timely for organisations to at least develop a strategy for how they will adopt and use Generative AI within their organisation (and keep up with future updates). At BDO, we have worked with other organisations to develop the following recommended approach to developing Generative AI strategies:

1. All-staff presentations on Generative AI to provide a level of education and awareness - we have had extensive feedback from our clients that this workshop helped demystify a lot of the uncertainties, reduce anxiety among team members and generate a level of excitement for its use within the organisation.
2. A leadership group level workshop – the aim of this workshop is to establish the culture and overarching direction that the organisation wishes to take in relation to Generative AI.
3. A management or working group level workshop – following the direction provided by the leadership group, this step involves developing a detailed roadmap for the implementation and adoption of Generative AI for the organisation. This involves exploring potential use cases, governance, security and funding requirements, the processes and people who need to be involved.
4. Technical implementation of a secure instance of Generative AI and application development for general use – This involves the technical implementation of a secure, private instance of Generative AI within the organisation that is available for general use by staff.

Whether it's data management, automation, adoption and collaboration, or machine learning and predictive analytics, BDO has AI solutions to assist you. Contact us for more information on the ever-evolving AI space.