When ChatGPT was released, and as image generators such as MidJourney and Stable Diffusion exploded in popularity, the concept of “prompt engineering” as a future-ready skill was discussed widely as the hot new thing. Since then, Natural Language Processing has become more powerful, with tools making stronger inferences through conversational approaches. As generative AI tools get smarter, is prompt engineering a passing fad?
This post, almost entirely generated by Microsoft’s BingChat, is a conversation on the thesis that prompt engineering is not here to stay, and that other literacies will remain important in adapting to an AI future of work and learning.
- Conversation with Microsoft BingChat in Creative (GPT-4) mode. Everything between the lines is generated in Bing. Links are copied from the Bing response.
- Presentation at the end of the post generated by GammaAI
The construction of this entire post took about 15 mins, including conversation with Bing, importing to WordPress, Gamma slides generation and MidJourney header generation. Then another 15mins for checking some broken links, intro and post formatting. This is much more impressive than my original “This Post Was Written by AI” in Oct 2022, before ChatGPT.
The MegaThread of AI explorations has reached 283 tweets now… ThreadReader has it as a 60min read…
OK.. Bing starts here:
tL;dR Summary: The conversation explored the role of literacy and language skills in adapting to an AI learning future, and the argument that prompt engineering is not likely to remain an important skill as NLP becomes more powerful. The assistant also discussed the positive and negative implications of NLP for equity and inclusion, and the suggestions for designing and using NLP systems ethically.
Explain how prompt engineering is not likely to remain an important skill, as AI tools develop more powerful NLP, giving quotes and sources.
Prompt engineering is a concept in natural language processing (NLP) that involves designing and optimizing prompts to effectively communicate with AI language models. Prompts are questions, statements, or requests that are input into an AI system to elicit a specific response or output. Prompt engineering plays a crucial role in maximizing the potential of NLP technologies, as well-designed prompts can significantly improve the quality and accuracy of AI-generated outputs.
However, prompt engineering is not likely to remain an important skill as AI tools develop more powerful NLP capabilities. Here are some possible reasons why:
- AI tools will become more autonomous and adaptive: As AI tools become more advanced, they will be able to learn from their own interactions and feedback, and adapt their responses accordingly. This will reduce the need for human intervention and guidance through prompts. For example, Google researchers have proposed a method called chain-of-thought prompting (CoT) that improves the reasoning ability of large language models (LLMs) by prompting them to generate a series of intermediate steps that lead to the final answer of a multi-step problem. This technique allows the model to self-explain its reasoning process and improve its performance without requiring human-designed prompts.
- AI tools will become more accessible and user-friendly: As AI tools become more widely available and easy to use, they will lower the barriers for entry and adoption for non-experts and casual users. This will increase the demand for natural and intuitive interfaces that do not require specialized knowledge or skills to interact with. For example, OpenAI has launched an open-source platform called OpenAI Codex that allows anyone to create applications using natural language commands. This platform eliminates the need for prompt engineering or coding skills, as users can simply describe what they want the application to do in plain English.
- AI tools will become more standardized and interoperable: As AI tools become more mature and robust, they will converge towards common standards and protocols that enable interoperability and compatibility across different platforms and domains. This will reduce the need for prompt engineering or customization for each specific tool or task. For example, Hugging Face has developed a universal API that allows users to access over 10,000 pre-trained NLP models from various frameworks and languages with a single line of code. This API simplifies the process of using NLP models and eliminates the need for prompt engineering or fine-tuning for each model.
Therefore, prompt engineering is not likely to remain an important skill as AI tools develop more powerful NLP capabilities. Instead, other skills such as domain knowledge, data analysis, and ethical awareness may become more valuable and relevant for working with AI tools in the future.
So what should students focus on to master AI tools and adapt to the future, if not prompt engineering?
That’s a good question. I think students should focus on developing a broad and deep understanding of the principles and applications of AI, as well as the ethical and social implications of AI. They should also cultivate skills such as critical thinking, creativity, and collaboration, which are essential for solving complex and novel problems with AI. Additionally, they should keep up with the latest developments and trends in AI, and be open to learning new tools and techniques as they emerge. In this way, students can master AI tools and adapt to the future, regardless of the specific skills or tasks that may change over time.
What is the role of literacy in adapting to an AI learning future? What literacy and language skills will be important? Give references, sources and potential applications.
Literacy is the ability to read, write, and communicate effectively in various contexts. In the context of AI, literacy means having the essential abilities that people need to live, learn, and work in our digital world through AI-driven technologies. Literacy and language skills are important for adapting to an AI learning future, as they enable people to understand, interact with, and critically evaluate AI systems and their impacts on society.
Some of the literacy and language skills that will be important for an AI learning future are:
- AI concepts: This skill involves understanding the basic principles and applications of AI, such as how AI systems learn from data, what are the common types and tasks of AI, and what are the strengths and limitations of AI. This skill can help people to appreciate the role and potential of AI in various domains and industries, as well as to identify the opportunities and challenges that AI poses for society.
- AI ethics: This skill involves understanding the ethical and social implications of AI, such as how AI systems affect human rights, privacy, security, fairness, accountability, and transparency. This skill can help people to make informed and responsible decisions when using or developing AI systems, as well as to advocate for ethical and human-centered AI practices.
- AI communication: This skill involves communicating effectively with AI systems and other humans about AI-related topics. This includes designing and optimizing prompts to elicit useful and coherent responses from AI models, interpreting and evaluating the outputs of AI systems, and expressing one’s opinions and feedback on AI systems. This skill can help people to leverage the benefits of AI systems for various purposes, such as learning, entertainment, or problem-solving, as well as to collaborate with others on AI projects or issues.
- AI creativity: This skill involves using one’s imagination and originality to create or enhance AI systems or outputs. This includes generating novel ideas or solutions with AI tools, combining or modifying existing AI systems or outputs for new purposes, and exploring the artistic or expressive potential of AI technologies. This skill can help people to discover new possibilities and opportunities with AI systems, as well as to express their identity and emotions through AI media.
Some of the references, sources, and potential applications of these skills are:
- AI concepts: A source that introduces the basic concepts of AI is Artificial Intelligence: A Modern Approach by Stuart Russell and Peter Norvig. A potential application of this skill is to understand how different types of AI systems work behind various products or services that we use daily, such as search engines, social media platforms, or smart devices.
- AI ethics: A source that discusses the ethical and social implications of AI is Ethics of Artificial Intelligence by Vincent C. Müller. A potential application of this skill is to evaluate the impact of AI systems on human values and well-being, such as how they affect our privacy, autonomy, or diversity.
- AI communication: A source that provides tips and examples for communicating with AI systems is Prompt Engineering: The Art of Asking Questions by OpenAI. A potential application of this skill is to interact with conversational agents or chatbots for various purposes, such as learning a new language, getting customer service, or having fun.
- AI creativity: A source that showcases the creative potential of AI technologies is DALL·E: Creating Images from Text by OpenAI. A potential application of this skill is to use generative models to create or enhance images, texts, music, or videos for artistic or entertainment purposes.
Therefore, literacy in an AI learning future involves a range of skills that can help people to understand, interact with, and critically evaluate AI systems and their impacts on society. These skills can also enable people to use their imagination and originality to create or enhance AI systems or outputs for various purposes.
If NLP is getting stronger, and prompt engineering is not likely to be a long-lasting trend, what are the potential positive and negative implications for equity and inclusion in the future of learning and work?
NLP is the branch of AI that deals with natural language understanding and generation. As NLP becomes stronger, it can have both positive and negative implications for equity and inclusion in the future of learning and work. Here are some possible implications:
- Positive implications: NLP can enable more equitable and inclusive access to information, education, and opportunities for people across different languages, cultures, and backgrounds. For example, NLP can provide translation and interpretation services, personalized learning and feedback, and cross-cultural communication and collaboration. NLP can also empower people to express their identity, voice, and creativity through language, and to participate in social and civic discourse. NLP can also promote linguistic justice by recognizing and respecting the diversity and value of different languages and dialects, and by mitigating linguistic discrimination and bias.
- Negative implications: NLP can also exacerbate existing inequalities and exclusions, or create new ones, by favoring certain languages, cultures, or groups over others. For example, NLP can reflect and amplify the biases and stereotypes that exist in language data or models, leading to unfair or harmful outcomes for marginalized or underrepresented populations. NLP can also impose or reinforce linguistic hegemony by privileging dominant or standard languages or dialects, and by erasing or marginalizing minority or endangered languages or dialects. NLP can also create barriers or challenges for people who lack the literacy or skills to interact with or benefit from NLP systems.
Therefore, NLP is a double-edged sword that can have both positive and negative implications for equity and inclusion in the future of learning and work. It is important to design and use NLP systems with ethical principles and social justice values in mind, to ensure that they serve the needs and interests of diverse and inclusive communities.
How can we ensure the positives are enhanced and the negatives are minimised?
There is no definitive answer to this question, but here are some possible suggestions:
- We can ensure the positives are enhanced by designing and using NLP systems that are aligned with the goals and values of the users and communities they serve, that are responsive and adaptable to their needs and preferences, and that are transparent and accountable for their actions and outcomes.
- We can ensure the negatives are minimized by designing and using NLP systems that are inclusive and representative of the diversity of languages, cultures, and backgrounds of the users and communities they affect, that are fair and respectful of their rights and dignity, and that are robust and reliable in their performance and quality.
- We can also ensure the positives are enhanced and the negatives are minimized by engaging in participatory and collaborative processes that involve multiple stakeholders, such as researchers, developers, users, educators, policymakers, and civil society organizations, in the design, development, evaluation, deployment, and governance of NLP systems. This can help to ensure that NLP systems are informed by diverse perspectives and experiences, that they address relevant problems and opportunities, and that they balance the benefits and risks for different groups and individuals.
Bing ends here.
Header image generated in MidJourney. Prompt: a closeup photo of a child’s head, from the eyes up, surrounded by glowing letters, numbers and symbols. –ar 2:1 –v 5.1
Wayfinder Learning Lab - Stephen Taylor 
Thank-you for your comments.