{"id":2838486,"date":"2023-01-18T14:02:23","date_gmt":"2023-01-18T19:02:23","guid":{"rendered":"https:\/\/onthinktanks.org\/articles\/\/"},"modified":"2023-02-24T15:18:35","modified_gmt":"2023-02-24T20:18:35","slug":"chatgpt-a-game-changer-for-the-knowledge-industry","status":"publish","type":"post","link":"https:\/\/onthinktanks.org\/articles\/chatgpt-a-game-changer-for-the-knowledge-industry\/","title":{"rendered":"ChatGPT: a game changer for the knowledge industry?"},"content":{"rendered":"
Like millions around the world, I spent the holiday period playing with ChatGPT<\/a> .<\/p>\n It is a large generative pretrained transformer (GPT) language model developed by OpenAI. It is trained to generate human-like text using machine learning algorithms that analyse an enormous amount of data and learn the patterns and structures of the language. In other words, it works by predicting the next best word in a series of words.<\/p>\n I started with a simple, and obvious, question: What is a think tank?<\/p>\n \u2018A think tank is an organisation that conducts research and provides analysis and policy recommendations on a variety of issues, often related to public policy, economics, and political science. Think tanks can be independent or affiliated with specific political parties or ideologies, and they often work to influence public policy and decision-making through their research and recommendations. Think tanks may also engage in public outreach and education efforts, such as hosting events or publishing reports or articles. Some well-known think tanks include the Brookings Institution, the Heritage Foundation, and the Cato Institute.\u2019<\/em><\/p>\n That is a pretty good definition, if you ask me. It picks up on think tanks\u2019 affiliations (e.g. to political parties) and their multiple functions (e.g. education).<\/p>\n Immediately I tried a second question: What is evidence-informed policy?<\/p>\n \u2018Evidence-informed policy refers to the use of research evidence, along with other forms of knowledge and expertise, to inform the development and implementation of policies and programmes.\u2019<\/em> Read more +<\/span><\/span> The full answer continued: ‘The goal of evidence-informed policy is to use the best available evidence to inform decision-making, in order to improve the effectiveness and efficiency of policies and programmes and to achieve desired outcomes.<\/p>\n The process of evidence-informed policy typically involves the identification and synthesis of relevant research evidence, the analysis of the strengths and limitations of the evidence, and the application of the evidence to inform policy decisions. It also involves engaging with various stakeholders, including policymakers, practitioners, and the general public, to ensure that research evidence is considered in a transparent and inclusive manner.<\/p>\n Evidence-informed policy is an important approach to policymaking, as it helps to ensure that policies and programs are based on the best available evidence and are more likely to achieve their intended goals. It is used in a variety of fields, including health care, education, social policy, and environmental policy, among others.’<\/p>\n This is also rather good. The definition considers \u2018other forms of knowledge and expertise\u2019, mentions the \u2018best available evidence\u2019 and includes engagement and inclusivity! I would have liked a reference to politics, ideology and ethics; but most evidence-informed policy practitioners wouldn\u2019t mention them either.<\/p>\n Then, at a Christmas party, a friend who works for a US Government contractor told me he\u2019d be using it in a much more sophisticated way, instructing it to produce original content. He felt that ChatGPT could present a real challenge to our practice.<\/p>\n So, there and then (in the middle of the party) I opened ChatGPT on my phone and jumped into the rabbit hole.<\/p>\n This article describes my first reactions and thoughts about how technologies such as ChatGPT may impact think tanks, the evidence-informed policy field and the kind of consulting that OTT does. I\u2019m sure these ideas will evolve as we learn more. And I look forward to hearing the reactions, ideas and fears of others too.<\/p>\n I must confess that, at first, the results I got from ChatGPT concerned me. But they also excited me.<\/p>\n This strategy document for a new (fictional) think tank in East Africa<\/a> is eerily good. The strategy was pretty good but it would need some detail and context to make it implementable.<\/p>\n The first outline for the evidence-informed policymaking course<\/a> was almost as good as anything I would expect from a course delivered by a leading actor in this field. Of course, the outline was not perfect. It missed modules on power, ethics and personal skills – although when I suggested them ChatGPT incorporated them.<\/p>\n Next, I asked ChatGPT to come up with a definition and examples of knowledge translation in the Global South.<\/a> We have been involved in a year-long research project on this topic for IDRC and I had the sinking feeling that ChatGPT would outdo us.<\/p>\n The definition and the examples (both fictional and real) were rather good. But when I asked ChatGPT to offer an analysis of the differences between knowledge translation in the \u2018North\u2019 and \u2018South\u2019 it struggled. It wasn\u2019t able to think outside the box and consider, for instance, that much of what is written about both knowledge translation and international development is written from the perspective of the global North and is therefore biased to a view of the world in which the South is unique and homogenous; both very wrong assumptions.<\/p>\n I tried asking it to write me an article on the potential impact of artificial intelligence on think tanks<\/a>. I found the result satisfactory content wise, but not style wise. I\u2019m sure that better instructions could improve the style.<\/p>\nMy first exploration of ChatGPT<\/strong><\/h1>\n