Artificial Intelligence (AI) systems like ChatGPT and Large Language Models (LLMs) are revolutionising the way that we work. These advanced models can generate human-like text responses, offering numerous benefits for startups. From automated customer support to creative content generation, language translation, educational assistance, and research support. They have the potential to streamline processes and enhance many different aspects of human-computer interaction.
In recent roundtable discussions hosted by Mandashi and Cambridge Angels, our portfolio companies shed light on 10 innovative ways in which they are using ChatGPT and the broader language AI revolution to accelerate their businesses:
Speeding up day to day processes
Startups were using ChatGPT and other LLM’s to streamline their day-to-day operations like writing blogs, performing research, and undertaking communications. ChatGPT was excellent at providing a starting point for developing new ideas, and circumnavigating the notorious “blank sheet problem”.
Many startups were looking at how ChatGPT could be used to enhance their products. For instance, it can provide a more intuitive interface for non-technical individuals to interact with complex data sets. An exciting application discussed was the use of ChatGPT to enable shop floor workers to query production line data and processes, leading to more efficient methods of working.
Enhancing Interactions and Validation:
By requesting ChatGPT to justify its responses and suggest methods for validating questions, startups found they could derive greater value from the AI system. Many startups are now developing strategies and prompts that elicit the most accurate and informative information from ChatGPT, ensuring more accurate and reliable interactions.
Deeper Interactions and Specific Data Sets:
As user expectations of Chat like interfaces grow, startups are exploring deeper interactions between ChatGPT and their own domain-specific data and products. The question was asked whether this will be best achieved by enhancing large language models (LLMs) to understand specific data sets, such as medical data, or whether smaller more focused models could be used instead.
Following on from this, many startups were interested in developing customised variants of ChatGPT tailored for their specific products. Intelligent integration tools were highlighted as an example, capable of interpreting informal text and transforming it into relevant configuration commands within an integration platform, significantly streamlining integration workflows.
In the healthcare sector, ChatGPT has demonstrated its effectiveness in answering medical questions. For example, a panel of licensed healthcare professionals preferred ChatGPT's responses 79% of the time and rated ChatGPT's responses as higher quality and more empathetic. Additionally, ChatGPT can be used to simulate patient behaviour, offering valuable insights for analysis. For instance, asking it to imagine being a patient with a specific condition. Of course, we have to be careful about hallucination and errors in diagnosis from the LLMs they are using.
Startups can leverage ChatGPT to assist with regulatory compliance tasks. By providing the relevant technical details, the AI system can generate relevant information required for completing compliance documentation. For example, this could be used for framing medical regulatory responses or completing security questionnaires.
Condensing Large Texts:
ChatGPT can condense large bodies of text, making them easier to digest and comprehend. However, it's important to note that accuracy may decrease if the system exceeds its text input token limit. As such, monitoring the output for accuracy is crucial.
Presenting Technical Information:
Startups can benefit from ChatGPT's ability to present technical and scientific information in a more accessible and business-oriented manner. This can enhance communication with various stakeholders, including patients and customers. An example was given of using ChatGPT to summarise complex medical terminology relating to rare cancers in a way that is more relevant and accessible to non experts. However, it was imperative to validate and cross-check results before publication. It was suggested that tools like Bing GPT can be used to provide references that help verify the accuracy of Chat GPT's responses.
ChatGPT proves to be an excellent tool for coding, even in less well-known languages. It understands code functionality and can ably assist with documentation, unit tests, and error message explanations. It can also be asked to offer multiple solutions, aiding developers in understanding different approaches to coding problems. However, caution should be exercised to prevent the introduction of unnecessary complexity or potential bugs.
Harnessing the Power of ChatGPT - the pros and cons
It's clear that most of the portfolio’s companies are excited about the huge potential that ChatGPT and LLM’s offer startups for various applications, and many are already making use of it in their business. However, it's important to acknowledge its potential risks and limitations. While it excels at handling clearly defined questions, open-ended inquiries may lead to misleading or unclear responses. Understanding the model's reasoning can be challenging, leaving room for false information to wrongly influence outcomes. It is consequently crucial not to blindly copy the generated output.
The use of AI systems like ChatGPT also raised concerns about data privacy and usage. An example is the handling of electronic patient data, where certification and regulation play crucial roles. Startups needed to carefully consider data retention and usage, especially when involving cross-border data transfers. It was proposed that more transparent options for opting out should be provided to ensure user trust.
Startups have an opportunity to leverage the power of ChatGPT and new advances in AI to enhance their products, streamline processes, and improve communication. However, it is important to be mindful of the strengths and limitations of AI systems, so that startups can harness their potential effectively. With careful validation and thoughtful application, it was agreed by all that startups have a great opportunity to unlock the true value of AI in their entrepreneurial journeys.
Cambridge Angels is a group of more than 60 high-net worth investors who have proven experience as successful entrepreneurs. Members invest in and mentor high quality start-up and early-stage companies in Cambridge, London, Oxford and throughout the UK.
Mandashi is a family investment company which invests in early stage startups and innovation that makes a positive difference to our world and the local Sheffield community.