New report from Transforma Insights finds that ChatGPT is a very sophisticated chatbot but wider enterprise application is likely to be limited.
ChatGPT has recently hit the headlines with the launch of prototype services in November 2022, a rumoured USD10 billion investment from Microsoft in January 2023 and the launch of enhanced services in March 2023. The March 2023 release is a significant improvement on that launched in November 2022 and, for instance, it is claimed that the latest engine scores in the 90th percentile in exams for entry into the US legal profession, whereas the version released in November 2022 scores in the 10th percentile. The latest version of ChatGPT can also describe images and communicate their meaning and context.
Transforma Insights has recently published a report 'ChatGPT is a significant step forwards, but has limited application in the context of enterprise digital transformation' analysing how ChatGPT might be relevant to enterprise digital transformation. As part of our analysis, we identified four different contexts for the deployment of ChatGPT against the backdrop of our full AI forecasts, including:
The number of AI instances associated with these four contexts is illustrated below. Two categories clearly dominate, including where ChatGPT is ‘Core’ or ‘Not Relevant’. This is largely due to the overwhelming dominance of AI instances that are associated with (i.e. on-board) IoT devices in the AI forecasts. In the scenario where AI Use Cases are deployed onto IoT assets, then each of those Use Cases will generally be critical to the functioning of the device, and there is little capacity to deploy solutions such as ChatGPT if they are merely ‘Relevant’ or only have ‘Tangential’ benefit.
Transforma Insights forecasts that by 2032 there will be 11.7 billion instances of AI deployed in contexts where a proposition such as ChatGPT could either potentially underpin or significantly enhance the solution, not including potential deployments associated with mobile or PC applications, other than where such applications are part of an enterprise process. These instances are dominated by Natural Language Processing, Chatbots & Digital Assistance and Machine Translation AI Use Cases deployed onto IoT assets including AV Equipment (such as televisions), Vehicle Head Units, and Personal Portable Electronics.
AI deployment scenarios for which ChatGPT has little relevance include situations in which definite and auditable feedback is required. For instance, ChatGPT doesn’t have the transparency (and is unlikely to have the robustness) required to underpin Fraud Detection, Predictive Maintenance or Voice Authentication. However, that is not to say that ChatGPT is completely irrelevant in some of these contexts, for example outputs from more basic (and less critical) Customer Behaviour Analysis, which ChatGPT can support, can be used as input to a Fraud Detection system.
Commenting on the findings of the research, Jim Morrish said: "Our analyses suggest that ChatGPT is essentially a very sophisticated chatbot with little relevance to wider enterprise digital transformation, particularly in contexts where AI must be robust, transparent and auditable. We think that this broad assessment is unlikely to change as ChatGPT evolves. In much the same way as we all use Google search, internet access, and mobile telephony (all of which are essentially ‘best efforts’) the ChatGPT solution may well soon be ‘good enough’ to rely on for many enterprise digital transformation purposes."
However, as discussed in the report, with new AI regulations envisaged in both the USA and Europe this may not be enough. Both sets of regulations focus on ‘explainability’ and some concept of risk management that takes into account the transparency of supporting AI engines, and ChatGPT does not score well for either explainability or transparency.
The quantitative assessment of the impact of ChatGPT is based on the Transforma Insights AI Forecasts, which were released in May 2022. See the press release 'New report from Transforma Insights predicts ten-fold growth in AI use over the next decade' for more details.
The AI forecast takes a unique perspective on the AI market, looking at the number instances of AI deployment, i.e. the count of independent machines (real or virtual) running machine learning algorithms of a particular type. It covers 42 use cases including Autonomous Systems, Image Processing & Analysis, Natural Language Processing, Repetitive Process Automation, Risk Analysis and System Optimisation. It provides splits by deployment type (cloud, edge, IoT and PCs/tablets/handsets), by country and by vertical sector.
All data is accessible via our AI Forecast Tool, as illustrated below.
Please contact us for more details.
