top of page

A Crash Course on Copilot & AI (Part One)


The world of Artificial Intelligence is rapidly developing at a rate faster than most people truly understand.


As the science evolves so does the numerous products being marketed by major tech companies in all industries.


Regardless of what role you play in your organization and whether or not you decide to incorporate AI into your work, it is essential that there is a basic understanding on what AI is and what it isn't.


We won't go over all the cool things that AI can do but will aim to provide the foundational knowledge of how it all comes together. Our goal is that by the end of this post, you feel better equipped to explore and discuss AI in your personal and work lives.


Let's get started


"AI" is an overused term. It's not necessarily that your computer is actually thinking (at least not yet), it's moreso just really smart at calculating things out so quickly that it feels like it is thinking.


Let's go with Microsoft's Definition from their Five Pillars of AI Leader's Guide to kick things off:


Artificial Intelligence (AI) (1950s): the theory and development of computer systems that are able to perform tasks that normally require human intelligence, such as visual perception, speech recognition, decision-making, and translation between languages
Machine Learning (1990s): a subset of AI and computer science where algorithmic models are trained to learn from existing data to make decisions or predictions
Deep Learning (2010s): a machine learning technique that uses layers of neural networks to process data and make decisions
Generative AI (2020s): a type of AI technology that uses algorithmic models to create new written, visual, and auditory content when given prompts or existing data


Today, when you hearing about or using "AI", typically this is referring to "Generative AI". Generative AI uses a "Model" which is basically a software that has been trained to simulate human style thinking.


A GPT is a model that has been designed to understand how we speak and communicate and then create responses based on that information.

If you have used something like ChatGPT or Copilot for Web, you are using a "Generative Pre-trained Transformer" aka a "GPT". It works by telling it something (entering a "prompt") and then it responds based on the information it has access to.


A GPT is a model that has been designed to understand how we speak and communicate and then create responses based on that information. This breakthrough has led to this technology being more exciting than ever.


Just remember, as smart as GPT is, it is not true "Artificial Generative Intelligence" (AGI). Meaning, it can't actually think for itself. All models today are based on a process called "training" where the model is given extraordinary amounts of data to learn from.


So why is this technology so exciting?


These new models fall into a couple of categories. If you have been following AI, there are all sorts of different model types that you might have heard of. A few examples might be Large Language Models (LLM), Large Action Models (LAM), or Multimodal Models.


Regardless of the type of model, the excitement comes from the ability for the model to analyze, compare, create, and work faster than anything that has existed before.


Some examples:


  • A recently released "ChatGPT-4o" from OpenAI is able to have fast and incredible sounding voice chats with you

  • Copilot for Web can gather information on the internet

  • Copilot for 365 can analyze your calendar and help you prepare for your meetings

  • Numerous image models can create brand new art and photos based on your description of what you want to see

  • Notetaking models can listen in on meetings and provide notes, ideas, and action items to organize your conversation

  • Writing models can help draft complete email and text messages

  • Visual models can "see" through a camera and interpret the world around it

  • Audio generation models can create brand new music, with lyrics, completely based on your description of what you want to hear

  • Multiple vendors have models that can read documents, summarize them, and answer questions about the document


In action, the technology is just impressive! There are tons of demos and videos on the internet to learn more. There's even this nifty blog post from a great IT company that goes over examples.


And that's just what exists today. Almost every major technology on the planet is investing heavily into AI. For many reasons, the results of these investments will dictate who dominates the market as AI solutions mature and become viable for the public to utilize.


It's scary too, so be cautious!


There are many new problems that AI existing creates. These problems require each individual and organization to pause, think out, and strategize their approach to incorporating AI into their life.


On one hand, you can ignore it completely, but more than likely it's coming whether you want it or not.


Here's a list of concerns about AI that have been highlighted by experts:


1. **Automation-spurred Job Loss**: The potential for AI to automate tasks and lead to unemployment¹.
2. **Deepfakes**: The creation of convincing fake audio or video content that can be used for misinformation¹.
3. **Privacy Violations**: AI's ability to collect and analyze vast amounts of personal data¹.
4. **Algorithmic Bias**: Decisions made by AI could be biased if the data it was trained on is biased¹.
5. **Socioeconomic Inequality**: AI advancements could widen the gap between socioeconomic classes¹.
6. **Market Volatility**: AI-driven trading and decision-making could lead to unpredictable market behavior¹.
7. **Weapons Automatization**: The use of AI in autonomous weapons systems raises ethical concerns¹.
8. **Uncontrollable Self-aware AI**: The hypothetical scenario where AI becomes self-aware and uncontrollable¹.
9. **Disinformation**: AI's role in spreading false information rapidly².
10. **Safety and Security**: Ensuring AI systems are safe from cyber threats and cannot cause harm².
11. **The Black Box Problem**: Difficulty in understanding how AI makes certain decisions².
12. **Ethical Concerns**: Ensuring AI is used in ways that align with human values and ethics².
13. **Bias**: The risk of perpetuating human biases in AI decision-making².
14. **Instability**: AI systems might behave unpredictably in complex environments².
15. **Hallucinations in LLMs**: Large language models sometimes generate nonsensical or unrelated content².
16. **Unknown Unknowns**: The potential for unforeseen consequences of AI actions².
17. **Environmental Impact**: The significant energy consumption required to train large AI models².
18. **Industry Concentration**: The risk of a few companies dominating AI development and use².
19. **State Overreach**: The potential for governments to misuse AI for surveillance and control².
20. **Role of Human Judgment**: Balancing AI's decision-making role with the need for human oversight³.
12 Dangers of Artificial Intelligence (AI) | Built In. 
https://builtin.com/artificial-intelligence/risks-of-artificial-intelligence
AI’s Trust Problem - Harvard Business Review. 
https://hbr.org/2024/05/ais-trust-problem

So, after all that, why are we still pushing forward with AI?


Well, what it all comes down to is... it's really awesome. Much like technology advancements in the past, this is a technology that will eventually alter how we work, play, and live our lives.


That said, as an individual, an employee, leader, and/or organization, it's important to be responsible and careful as you move forward with AI.


Some quick notes:


It is time to be curious but remain skeptical!

  • There are many companies adding "AI" to their products that might not actually be AI

  • Trusting AI is possible, but the need to verify will continue to be important

  • Watch out for random companies marketing AI solutions to you!

  • Everything that exists today will be nothing compared to where we are in a year or two, things are moving too quick!

AI is expensive (for now)

  • There are reasons that chipmakers like Nvidia are setting new profit records. People need processors that are AI capable and there are only so many resources available to make them!

  • Companies and Governments are buying very expensive chips to use for model training and AI development, so expect there to be a cost with any AI that you use!

  • As models and processors evolve, new types of AI processing is becoming available that is more efficient and more readily available

  • "Free" is not Free. If you are using a "Free" GPT, your data is likely the fee.

Focus on privacy and the security of your data

  • Unless you are paying for a service (and even some of the paid services do this to) your data may be used to train models. This can create severe security and privacy issues, especially for organizations that need to protect sensitive information.

  • Remember that AI is very fast.

  • If I gave you 1000 pages filled with codes for Amazon gift cards and said you could enter as many as you can in 5 minutes, you'd barely get past page 1.

  • AI can memorize all 1000 pages in seconds, but it might also be able to enter all of the codes and make recommendations and budgets on how to spend it all!

  • Imagine it had access to your personal information or your company databases. There's a lot of good (and bad) that can happen very quickly!


AI Vendors and Products are plentiful


There are a lot of "AI" products that exist. There are several vendors that are incorporating "AI" into their product. Some might be developing their own models specifically for their product.


However, most commercial offerings are tapping into the power of the major AI models owned by OpenAI or Microsoft.


  • OpenAI essentially is dominating the industry in research and development of AI and GPT models

  • Microsoft is a close partner and investor in OpenAI, and has their own iteration of OpenAI's models and technology that has been customized for business and consumer use


For the vendor, they are being charged by OpenAI or Microsoft to use their computing resources ("Compute") and then the vendor passes those costs to you via a usage license or subscription.


Other major vendors are working on their own models and solutions such as Google Gemini and Amazon Bedrock. Apple hasn't revealed their AI offerings yet, but that is expected to be announced soon (Likely using OpenAI's tech)


Then, there are smaller vendors and developers with their own models and solutions that are specializing on specific functions to be used in their own solutions.


From our perspective, as of today, Microsoft is well ahead of the pack and has created a foundation and platform that most of the industry is rallying around. If you haven't heard of the term "Copilot" get ready, cause the entire PC industry is planning a major marketing push.


Wrapping Up


If you made it this far, we commend you! AI will continue to be a hot topic in every industry and we hope this post helped provide clarity!


In Part 2, we'll focus on Microsoft's Copilot platform, Copilot+, and the recently announced "AI PC"s that launch in Summer 2024.


This is the first part of a two-part post, click here to read part two.



Interested in having a conversation about technology in your business? 

Send in a contact request or email info@aevotec.com.

Comments


Newsletter Sign-Up

Thanks for signing up for our newsletter!

bottom of page