The State of Artificial Intelligence: 2023 

By Greencastle Consulting

Alan Turing began artificial intelligence (AI) research in the 20th century but in 2023 AI was brought to the masses. 

The buzzword ‘artificial intelligence’ has appeared in news headlines, dinner conversations, blogs, social media, newsletters—anywhere people consume information. From doing research to writing emails to assisting in project implementations, the impact AI has had on our daily and professional lives is massive.   

In the past year, AI has made a significant impact on the world of project implementations. Anyone involved in a project can leverage AI models to assist in various aspects of their daily work. AI can automate administrative tasks, giving individuals more time to focus on larger aspects of the project. These models learn and become smarter every day, enhancing your ability to perform your job more efficiently. 

AI in 2023

Most people probably heard about OpenAI’s ChatGPT in 2023.  ChatGPT is an application built off a Large Language Model (LLM) that can take questions, in plain language, while returning results in a similar fashion.  Behind the scenes, LLMs use large advanced data science models called transformer model neural networks that are “trained” on an incredible amount of data.  

ChatGPT may have dominated the news in 2023, but neither the application itself nor the underlying models (such as GPT 3.5 and GPT4) are the sole players in the field.  Other companies, such as Anthrop\c (Claude) and Google (PaLM2 / Gemini), created models and applications based on them.  There are even open-source LLMs being created by the community. 

In 2023 AI tools were not restricted to just returning text; tools such as Midjourney and Dall-E are available to create images from text prompts.  Recent innovations in this space include using competing neural network models (Generative Adversarial Networks) that are increasing the quality of these AI-generated images. 

Retrieval Augmented Generators (RAGs) have extended AI models and applications. RAG methods allow developers to upload their own body of knowledge to an AI app and then use that new information to craft the best answer to a question.  Consider an HR assistant application, that uses an existing LLM model (like Meta’s LLaMA 2), but has no information about your company policies or procedures.  HR questions would yield generic results, or suggested best practices.  But with RAGs, that same LLaMA model can be enhanced with your company handbook and policy documents, giving the application better context that results in more accurate answers.  Vector databases, adept at storing complex data in a high-dimensional space, have greatly benefited the loading of these knowledge bases. 

The appeal of the latest AI is how many of these are deployed in low-code / no-code environments that make a difference for everyday people.  This generation of AI is available in virtually every industry and users don’t need specialized technical skills to get results. 


AI for Project Implementation 

The role of AI in project implementations is still evolving but understanding how to use these models in your own work is powerful. Effectively utilizing AI can bring about efficiency and automate some tasks that allow you to focus on other priorities. 

  • In the project management space, project plans and Work Breakdown Structures can be generated in seconds using AI saving valuable time. 
  • Change managers can leverage AI-generated summaries of new frameworks or create workflows to pass along to junior employees creating shared understanding. 
  • Process improvers can use AI to generate fishbone diagrams with a couple of sentences leading to a better understanding of the problem and ways to improve. 
  • Subject Matter Experts (SMEs) can build AI solutions fit for their organization, while data scientists and data engineers can focus on advanced use cases and emerging tech.  There are already new roles for Prompt Engineers, people who are experts in querying these LLMs. 

Overall, low-code / no-code systems have the potential to drive AI adoption across industries. 

AI for the Future 

The journey continues and with great power, comes great responsibility. It’s time to embrace the AI tool’s potential to be a force multiplier in your business and personal life.  Exploring these and future technologies will help foster innovation, save time, and drive results. 



AI Glossary of Terms 

Alan Turing: A British scientist and WWII Enigma code breaker who conceptualized the Turing Test for computer intelligence. 

Turing Test: A test proposed by Alan Turing in 1950 to evaluate a machine’s ability to exhibit human-like intelligence. 

Artificial Intelligence (AI): The simulation of human intelligence processes by machines, particularly computer systems. 

AI Research Winter: Periods when progress in AI research slows down or funding decreases, often due to unmet expectations. 

Neural Networks: Computational models inspired by the human brain, used in machine learning for pattern recognition and decision-making. 

Large Language Model (LLM): AI models trained on massive text data to acquire strong language understanding and generation capabilities. 

Transformer Model Neural Networks: Advanced neural network architecture well-suited to processing sequential data like text and speech based on self-attention. 

ChatGPT: Conversational AI system from OpenAI built using large language models that can understand questions and generate natural language responses. 

GPT-3.5 and GPT-4: Iterations of OpenAI’s Generative Pre-trained Transformer models, known for their advanced language processing capabilities. 

Generative Adversarial Networks (GANs): A class of machine learning frameworks where two neural networks compete to enhance the quality of outputs, often used in image generation. 

Retrieval Augmented Generators (RAGs): AI methods that enhance existing models with additional specific data, improving context and accuracy in responses. 

Vector Databases: Specialized high dimensional databases optimized for serving neural network models including recommendations. 

Low Code / No Code Environments: Platforms enabling non-technical users to leverage and customize AI through intuitive graphical interfaces instead of coding. 

Prompt Engineer: A role specializing in effectively querying Large Language Models to generate desired outputs. 

Hallucination in AI: A phenomenon where AI models generate false, misleading, or entirely fabricated information. 

Note: Generative AI was used to help craft the Executive Summary and the Glossary.