Generative AI: From prompt engineering to problem formulation


Generative AI(GA) is a branch of artificial intelligence that focuses on developing algorithms that can create new content. This content can be anything from text and images to music and code. Generative AI models are trained on large datasets of existing content, and they can then be used to generate new content in a variety of ways.

Prompt engineering is the process of designing prompts that can be used to guide Generative AI models to produce the desired output. A good prompt engineer is able to craft prompts that are clear, concise, and specific. They also need to have a good understanding of the capabilities and limitations of the Generative AI model they’re using.

Problem formulation is the process of defining a problem in a way that can be solved by a Generative AI model. A good problem formulator is able to break down a complex problem into smaller, more manageable subproblems. They also need to be able to identify the key data and inputs that are needed to solve the problem.

Generative AI is shifting from prompt engineering to problem formulation for a number of reasons. First, Generative AI models are becoming more powerful and capable. They are now able to solve problems that were once thought to be intractable. Second, the field of problem formulation is maturing. There is now a growing body of research on how to formulate problems in a way that can be solved by Generative AI models. Third, there is a growing demand for Generative AI solutions to real-world problems. Businesses and organizations are looking for ways to use Generative AI to improve their efficiency, productivity, and creativity.

Are You Ready to Ride the Generative AI Wave? Discover How!

Exciting times ahead in #GenerativeAI and #PromptEngineering! These fields are evolving rapidly, poised to revolutionize industries with their creative potential. Explore the latest trends, techniques, and applications. 🚀

Problem formulation and GAuse cases ?

Generative AI is increasingly being used to solve real-world problems in a variety of fields:

  • Drug Discovery: Generative AI models can generate new potential drug molecules, speeding up the discovery process and reducing costs.
  • Materials Science: Generative AI can be used to design new materials, optimize existing ones, or discover new properties in known materials.
  • Natural Language Processing (NLP): Generative AI can produce human-like text, making it useful in areas like chatbot development, content creation, and language translation.
  • Computer Vision: Generative AI models can generate realistic images, useful in fields like entertainment and security.

Generative AI is also being utilized in other sectors, including finance, entertainment, and transportation.

However, with these opportunities also come challenges. Bias in AI models can lead to unfair outcomes, while misuse of Generative AI can result in issues like deepfakes or spam content. On the positive side, the democratization of creativity and innovation through Generative AI can lead to a wide range of new ideas and solutions. Generative AI is making our lives easier and more efficient, and is being used to solve real-world problems across various industries.

problem formulation use case : A dynamic montage of different scenarios showing Generative AI in action. This could include a scientist in a lab coat examining a molecular structure, a writer at a computer generating content, and a security camera capturing a generated image. Overlay AI-related graphics and data visualizations to convey the breadth of AI applications in drug discovery, content creation, and computer vision.

How Generative AI models are trained to generate human-like text in natural language processing?

Generative AI models are trained to generate human-like text in natural language processing (NLP) using techniques from machine learning and deep learning. The process of training these models is often complex and involves several steps:

  1. Data Preprocessing: Before a model processes text for a specific task, the text often needs to be preprocessed to improve model performance or to turn words and characters into a format the model can understand. This can involve stemming and lemmatization (converting words to their base forms), sentence segmentation (breaking a large piece of text into linguistically meaningful sentence units), stop word removal (removing the most commonly occurring words that don’t add much information to the text), and tokenization (splitting text into individual words and word fragments).
  2. Feature Extraction: Numerical features extracted by the techniques described above can be fed into various models depending on the task at hand. For example, for classification, the output from the TF-IDF vectorizer could be provided to logistic regression, naive Bayes, decision trees, or gradient boosted trees. Or, for named entity recognition, we can use hidden Markov models along with n-grams. Deep neural networks typically work without using extracted features, although we can still use TF-IDF or Bag-of-Words features as an input
  3. Modeling: The model is then trained using a large dataset, such as a corpus of text from Wikipedia or another large source. The model learns to predict the next word in a sentence given the previous words, effectively learning the structure of the language. This is done by feeding the model a sequence of words and having it predict the next word in the sequence. The model’s predictions are compared to the actual next word, and the model’s parameters are adjusted to minimize the difference between the predicted and actual word. This process is repeated many times on many different sequences of words from the training corpus.
  4. Fine-tuning: Once the model has been trained on the large corpus, it can be fine-tuned for a specific task. This involves training the model on a smaller, task-specific dataset, allowing it to adapt its previously learned language structure to the specifics of the new task.

One of the most well-known examples of a generative AI model used in NLP is GPT-3, developed by OpenAI. GPT-3 uses a transformer architecture, which relies on a self-attention mechanism to draw global dependencies between input and output].

It’s important to note that while these generative AI models can produce remarkably human-like text, they don’t understand the text in the way humans do. They’re simply predicting what word is likely to come next based on their training data. Also, these models can sometimes generate text that is biased or otherwise problematic, as they can only mimic the patterns they’ve seen in their training data.

a key unlocking a treasure chest filled with diverse items representing the various fields Generative AI is impacting. The key could be shaped like a prompt, and the chest could contain images of drug molecules, a computer screen with text, materials, and more


In summary, Generative AI is a powerful tool that is moving from the realm of prompt engineering to problem formulation. This shift is allowing for more creative and flexible solutions to real-world problems, from drug discovery to new material design. However, it’s important to continue addressing the challenges of bias and misuse in this field.

Examples of Generative AI models

One of the most well-known examples of Generative AI is GPT-3, a language generation model developed by OpenAI. It can write essays, answer questions, and even create poetry.

In the field of drug discovery, models like Reinvent and LatentGAN have been used to generate novel drug-like molecules.

How to get started with Generative AI ?

To get started with Generative AI, one could begin by learning about machine learning and deep learning principles. Various online platforms offer courses in these areas. Then, one can start experimenting with Generative AI by using pre-trained models available in libraries like TensorFlow and PyTorch.

Resources for learning more about Generative AI

For further learning, resources like Google’s Machine Learning Crash Course, OpenAI’s documentation, and various research papers available on arXiv are great starting points. Additionally, online communities like Kaggle and StackOverflow can provide practical insights and help solve any challenges faced while working with Generative AI.