ChatGPT – Pushing the boundaries or killing the need for CI?
4 April 2023
For years, natural language processing (NLP) has fascinated computer scientists, and anyone concerned with the interface of linguistics and new technology1. But it is only recently, with OpenAI launching a freely accessible (at least in its basic version) language-processing AI model through GPT-3 and GPT-4 (Generative Pre-Trained Transformer-3 and -4), that it has been able to shake up the (chatbot) world. With its ability to understand human language, ChatGPT can distil information and respond to complex questions by learning from all text it has been fed. The concept of input and output seems very simple and is yet backed with a rather complex technology that continues to learn while guessing what the next word should be, constantly improving its understanding of prompts and questions to become the ultimate know-it-all.
The technology has been around for quite some time; big tech companies are all working on AI language generators in the background. Google is working on versions of LaMDA, Meta has Blenderbot, Baidu launched Ernie, and Microsoft Research is working on a BioGPT2 version that is supposed to help researchers gain new insights; for example, in drug development or clinical therapies. The sheer investments of these companies into NLP to enhance their existing products and services is only the beginning and gives an idea of what potential the technology might unleash to disrupt our way of working and living.
Is ChatGPT about to disrupt the wider consulting industry, market research, and competitive intelligence?
Before getting overly excited or instead too doomed about the future, it is undoubtedly prudent to get our heads around the impact on today’s business and how consulting – and, specifically, the competitive intelligence and market research offering – has an exciting opportunity to evolve in the future.
- The current language processing software can compress complicated matters to impressive yet often slightly superficial content. This can pose a threat to pure content-producing services.
- Language models are based on past data and, as such, (most probably) will have trouble generating truly innovative, future-focused solutions.
- For market research and competitive intelligence services, referring to a source is crucial and needs solid baselining to build (qualitatively and/or quantitatively robust) hypotheses. ChatGPT and other software are not trained and capable of validating the sources or making hypotheses effectively.
- Highly sensitive topics in a clinical context, based on the latest outcomes in research, the interpretation of information, and its implications, need to be analysed more cautiously than GPT can do – for now.
Tomorrow and in the further future
- ChatGPT offers a great head start and can build a basis for more in-depth analysis as well as hypothesis generation. Research-heavy project types, such as market studies and competitor deep dives can benefit from this.
- Policies will need to be put in place that bring transparency to the customer into the part of the consulting and research service that is being complemented with NLP software such as chatbots.
- Staff need guidance on when NLP software application is adequate and how it should be used.
- Critical thinking and the skills to question the tools used will be key and must lead to an evolution of knowledge management.
- The quality of the question will become important like never before….
Why should customers care?
Although AI or a chatbot may help to streamline a consultant’s workload, the tasks often involve solving complex, open-ended issues. These require a blend of ingenuity, scientific knowledge, connected thinking, intuition, and pragmatism – something the technology is not able to mimic yet. Critical thinking is crucial to understanding the impact of the incomplete or biased data that is being input and to identifying incorrect output, even if it is being presented with a high degree of confidence. Likewise, the provision of potential future scenarios and their assessment of the probability of these becoming a reality will continue to depend on human judgment.
Another essential aspect that should be remembered is that consultants also build rapport with clients and create strong business relationships by finding common ground and understanding the nuances of their current challenges – just like the interplay of medical technology, or a certain treatment, and a human doctor in providing care to a patient. The perceived uniqueness as individuals for the service or treatment we receive still plays an important role. An experienced consultant builds trust and can understand the implicit ask and investigate further connected aspects while putting them in context. These very human skills have yet to be developed, at least in mainstream AI systems.
Natural language processing software has great potential in many ways and will contribute to productivity. One can be the relief from time-consuming manual and monotonous data processing or the automation of basic research tasks with high-volume content. The way forward will be to grow both a company and its services together alongside AI technology as that becomes more available (see illustration 1), just as the service industry transformed from paper-based to digital.
When adopting these tools, it will be of utmost importance to upskill staff to classify the created output. What will also become much more crucial in the future will be enhancing the processes, identifying the ethical use of these tools and, equally importantly, critical thinking about the sources used and the context that comes with it.
The value that consultants continue to bring to the customers, and will professionalize further, is a thorough understanding of their problems to be solved and a greater awareness of their commercial context. CI experts will be able to uncover the hard-to-reach insights that others overlook. They will be able to understand the essential information about competitors that is available to the public and develop valid hypotheses on the drivers of the decision-making processes. Even more important, they will be able to put this into a perspective that is relevant to the customers’ business with all its nuances, because there is always more behind the explicit ask than formulated in the project scope. That is why we keep a big-picture analysis at the forefront of our engagements, cycling between broader topics to detailed questions in order to improve our probability of finding the unknowns. Exciting times lie ahead to provide even better analysis to the client, empowering them to make even better
The future is auspicious. Until the technology matures, we will continue to identify relevant use cases and thoughtfully test different means to make our operations smoother, faster, more precise and more efficient. We do not want to compromise on building knowledge and expertise, or ignore the importance of not trying, failing, and testing. Critical thinking will continue to be at the centre of our actions and services. Ultimately, we strive to ensure that our employees are equipped with the right tools to help relieve the burden of monotonous tasks, free up time for more exciting tasks and, simultaneously, increase their satisfaction and happiness at work. We are convinced that happy employees will lead to more satisfied clients and more tailored service delivery.
Graph inspired by Gartners graph here. Chatbots in use in consulting and research will most likely take longer than GARTNER’s graph indicates to first be implemented successfully and then to reach a plateau of productivity – especially considering the open letter of the future of life institute that calls for a pause of AI experiments that go beyond GPT-4.
1 The Association for NLP, February 2023
2 The Decoder, February 2023