Five development questions to ChatGPT

Arnaldo Pellini
6 min readJan 9, 2023
DeepMind on Unsplash

I listened before the Christmas break to an episode of One and Tooze, which had a segment about AI and the latest release by OpenAI’s ChatGPT. This is the bot launched by OpenAI, which had more than a million users within a week of its release.

Open AI explains that “ChatGPT is a language model trained to produce text. ChatGPT was optimized for dialogue by using Reinforcement Learning with Human Feedback (RLHF) — a method that uses human demonstrations to guide the model toward desired behavior.” They also add: “These models were trained on vast amounts of data from the internet written by humans, including conversations, so the responses it provides may sound human-like. It is important to keep in mind that this is a direct result of the system’s design (i.e. maximizing the similarity between outputs and the dataset the models were trained on) and that such outputs may be inaccurate, untruthful, and otherwise misleading at times.”

OpneAI warns that “ChatGPT is not connected to the internet, and it can occasionally produce incorrect answers. It has limited knowledge of the world and events after 2021 and may also occasionally produce harmful instructions or biased content. We’d recommend checking whether responses from the model are accurate or not. If you find an answer is incorrect, please provide that feedback by using the “Thumbs Down” button.”

I wanted to test it by asking five questions related to my research and works in development and you can read the Q&A below. I also asked ChatGPT to write me the starting paragraph for a blog about ChatGPT. Are is the answer it gave me:

Welcome to our blog about ChatGPT, a powerful language processing tool developed by OpenAI. ChatGPT is a variant of the GPT-3 model, which has been trained on a massive dataset of internet texts and can generate human-like text in a variety of languages. In this blog, we will delve into the capabilities and potential applications of ChatGPT, as well as examine some of the ethical considerations surrounding its use. Whether you are a developer looking to build chatbots or a language enthusiast looking to learn more about natural language processing, this blog has something for you.”

Here is the Q&A

Q1 — Give me a definition of a complex system

ChatGPT — A complex system is a system that is made up of many interconnected parts that work together in complex and unpredictable ways. These systems are often found in natural systems, such as the human body or the ecosystem, but can also be man-made, such as a computer network or a social organization. Complex systems are difficult to understand and predict because of the many factors that can influence their behavior.

Q2 — Define political economy analysis

ChatGPT — Political economy analysis is the study of the interaction between political and economic factors. It seeks to understand how political institutions, policies, and decisions affect the economy and vice versa. Political economy analysis often focuses on the ways in which power and influence are distributed within a society and how these dynamics affect economic decision-making and outcomes. This approach can be used to study a wide range of issues, including economic development, trade policy, and the distribution of wealth and income.

Q3 -Define a portfolio approach for social system shift

ChatGPT A portfolio approach to social system shift is a strategy that involves pursuing multiple, diverse, and complementary interventions simultaneously in order to bring about significant and lasting change in a social system. The idea is to create a balanced and diverse “portfolio” of interventions that can address different aspects of the problem and increase the likelihood of success. This approach recognizes that social systems are complex and dynamic, and that a one-size-fits-all solution is unlikely to be effective. Instead, it seeks to identify a range of interventions that can work together to bring about the desired change.

Q4 — What is adaptive development in international aid and cooperation

ChatGPT — Adaptive development in international aid and cooperation is a approach to development that emphasizes the need for flexibility and adaptability in the design and implementation of development programs. The idea is that development programs should be designed in a way that allows them to adapt to changing circumstances and to the needs of the communities they are intended to benefit. This approach recognizes that the complex and dynamic nature of development means that it is difficult to predict exactly what will work in any given situation, and that development programs must be able to adapt and evolve over time in order to be effective. As such, adaptive development often involves a focus on learning and evaluation, and on building the capacity of local communities and organizations to plan, implement, and adapt development programs.

Q5 — Define developmental evaluation

ChatGPT — Developmental evaluation is an approach to evaluation that is used to assess the effectiveness of complex, innovative, and adaptive programs and initiatives. This approach is often used in the context of development programs, where the goals and outcomes are difficult to predict and measure, and where the program is expected to evolve and adapt over time. Developmental evaluation focuses on supporting the ongoing learning and adaptation of the program, rather than simply assessing its success or failure. It typically involves a close collaboration between the evaluator and the program team, and a focus on generating and using real-time feedback to inform program design and implementation. This approach is often used in conjunction with other forms of evaluation, such as formative and summative evaluation, to provide a more complete picture of the program’s effectiveness.

The answers are pretty good. They could be helpful prompts for writing a blog, refreshing my memory about specific definitions, and as inputs to a paper.

While playing with it, I thought about how it could help my development research and consulting work. It is a knowledgeable assistant that can quickly provide answers and suggestions. I can use it when working on context analysis or political economy analysis. For example, I could ask ChatGPT to provide me with a summary of the political and historical development of a country. I could ask about the main barriers to scaling poverty reduction initiatives in a region or a specific nation. I could ask it to summarise the factors that have, for example, reduced the adoption of digital technology in education.

These can be prompts or suggestions for more focused analysis and collecting primary and secondary data. It is similar to a preliminary scan of a vast library of online documents and helps focus the research on some specific political economy insights.

ChatGPT will not provide literature and references to the text it produces. Instead, it may give the name of a researcher or an author. Therefore, the answers may read as general at times.

It is an assistant, a very knowledgeable assistant that we should not be afraid of. Instead, we should learn to work with it by recognising what it can do for us and what it cannot and acknowledging that we work with it.

It can help improve how we explore, research, and analyse contexts and social development problems in search of answers to help reduce poverty in the world.

Let me know what you think.

You can listen/read more about ChatGPT on The New York Times podcasts:

Did Artificial Intelligence Just Get Too Smart?

Can ChatGPT Make This Podcast?

And in The Guardian:

What is AI chatbot phenomenon ChatGPT and could it replace humans? AI bot ChatGPT stuns academics with essay-writing skills and usability The ChatGPT bot is causing panic now — but it’ll soon be as mundane a tool as Excel

Photo credit: DeepMind on Unsplash

Originally published at https://knowledgecounts.fi on January 9, 2023.

--

--

Arnaldo Pellini

Director of www.capability.fi • adaptive MEL systems • knowledge systems • problem-driven development • www.knowledgecounts.fi • own view