Our starter for 10 on GenAI

By Paul Graham and Bethany Blanes, Microsoft Copilot and Power Platforms Productivity Coaches, Prosource.it 

A recent study has shown that 75% of knowledge workers are already using GenAI and it is so useful to them that nearly half wouldn’t give it up even if it was banned by their employer.  

Undoubtably the ability of GenAI to help speed up or improve our work is very attractive, but for businesses it’s not as simple as just adopting it because it’s cool (or at least it shouldn’t be). There are significant risks if it isn’t adopted responsibly and ShadowAI, which is where employees simply make their own choices and use it anyway, is a definite danger. 

Properly understanding the needs of the business before assessing future benefits is an essential first step. But with so many providers available it can be hard to keep track of the differences, as well as the pros and cons of each. Recent news of the Chinese contender DeepSeek leaping into the fray of AI models means that this is yet another option. So if you don’t know your Claudes from your Geminis, and your head is spinning trying to understand the many acronyms buzzing around, then read on for our quick overview to make sense of it all. 

Get your definitions straight 

It can get confusing when different terms are being used interchangeably, so let’s start with some definitions. 

AI or Artificial Intelligence is a system that has been created to mimic human behaviour and decision making. It has wide ranging applications on everything from decision-making or planning through to specific uses such as video, translation and text generation.  

Generative AI, which we’re hearing a lot about in the media at the moment, is a specific type which is capable of generating new models using what has been learnt, and this can be used for text, imagery and music among other things. 

Large Language Models (LLMs) are a type of generative AI specifically focused on natural language processing (NLP). This can be applied to a wide range of uses where realistic human language is needed. 

So that’s the acronyms covered (at least for now….) 

A potted history of LLMs  

As much as it might seem like AI is something that arrived out of nowhere a couple of years ago, the truth - like any overnight success - is that it’s has been in development for many years. It all started back in the 1960s with something called ELIZA, basically a chatbot that tried its best to have simple conversations. By the 1990s, scientists figured out how to create language models that could guess what word might come next in a sentence. These models showed that with enough practice (or data), machines could learn to talk like us.  

Speech recognition software helped a lot, too. Back then, programs had to guess what word you were saying even if you mumbled! This “guessing game” turned out to be super useful for teaching LLMs to think ahead, like figuring out the next word in a text message.  

The major players in AI  

Now let’s check out the big names in the AI world and what they’re up to:  

• OpenAI’s GPT: Known for its versatility, ChatGPT excels in tasks ranging from content creation to coding assistance. Its integration into various platforms showcases its adaptability. 
• Microsoft Copilot: Imagine having a helper right inside your favourite apps like Word or PowerPoint. Copilot can whip up reports, make slides, and even summarise your emails. Perfect for staying on top of work or school projects. 
• Google’s Gemini: This one is all about teamwork. Gemini works well with Google’s other tools, like Google Cloud, making it a great choice for developers and businesses that are already in Google’s universe. 
• Anthropic’s Claude: Prioritising ethical AI usage, Claude focuses on safety and transparency, catering to industries that require responsible AI deployment. 
• xAI’s Grok: Grok is the new kid on the block. Made by Elon Musk’s team, it works with X (formerly Twitter). It’s still figuring out its place in the crowd, but it’s trying to catch up with the big players. 
• DeepSeek: For many, the first they had heard of this was in the last month when it suddenly topped Apple store’s downloads. Chinese owned and allegedly created for a fraction of the price of the industry leaders, this is the very latest entrant (but we’re sure not the last). 

Choosing the best option 

For businesses, choosing the right GenAI model involves assessing security, scalability, and integration capabilities. Models like Microsoft's Copilot offer deep integration with existing tools, while Claude's emphasis on ethical AI makes it suitable for sensitive applications.  

Like any new application, it’s about thinking about your business and your people, then embedding the new technology in the way that is best for you. 

One very important point to note is that the free versions of these applications will use everything shared with them to train their AI systems to be smarter. That’s fine for helping to plan your next holiday or for some meal ideas, but not for business-critical information. Many companies are updating their policies to make sure that staff are not using these open models in their work. The paid-for models provide closed systems which means that they will not use your information to train their models. Security is a critical consideration which we mustn’t lose sight of in our rush to try these shiny new toys. 

The big picture  

AI in all its guises is here to help us all be a little better at what we do whether that’s writing, solving problems, or just getting through a busy day. At prosource.it, we’re all about using these tools to make life easier and more exciting – supporting our clients to understand them better. Because when AI’s on your side, the possibilities are endless!