The Salford Innovation Forum recently witnessed an exceptional AI event showcasing the cutting-edge developments in artificial Intelligence.
Featuring a captivating guest speaker Dylan Humphreys who is a seasoned technology professional with 25 years experience in the ICT sector. Specialising in artificial intelligence, prompt engineering and ChatGPT. Going on to form Dixon Humphreys, industry leading experts who work with organisations across multiple sectors to develop AI capabilities, who provided invaluable insights into the world of AI-driven conversation.,
With the stage set to unravel the secrets of ChatGPT, the acclaimed language model, this event was a rare opportunity for attendees to gain a deeper understanding of how this technology can be harnessed to enhance various applications. In this article, we will explore the highlights of the event, with a particular focus on the enlightening presentation by our guest speaker, as they guided the audience through the practical usage of ChatGPT and its transformative potential in the realm of AI-powered conversations.
Introductions and the Recent History of AI
We embark on a journey of discovery, aiming to delve deeper into the fascinating world of artificial intelligence and its implications. As the event’s host, we clarified that while the AI-generated advertisement may have promised Rob, we are indeed the Humphreys of Dixon Humphreys, here to continue the exploration of AI’s transformative potential. With AI playing a significant role even in generating his bio and headshot, it’s evident that we’re living in an era where AI is rewriting the rules of engagement.
We understand that the key to harnessing AI’s potential lies in answering fundamental questions: What is AI, why is it so significant now, and how does it differ from previous AI developments?
We’ll also touch on the mechanics of AI and its real-world impact. Our journey begins with OpenAI, founded in 2015 as a nonprofit organisation with a mission to advance artificial general intelligence (AGI), akin to human-level reasoning. In 2018, OpenAI transitioned to a for-profit model, marking a shift in its approach. Fast forward to 2020, the launch of GPT-3, which laid the groundwork for advancements like Chat GPT, although initially requiring a fair degree of technical proficiency. In 2021, Dali, OpenAI’s image generation AI, joined the roster, and as we move into 2022, we witnessed the momentous launch of Chat GPT, making history by reaching 100 million users in just over two months, a feat that even Instagram took two and a half years to achieve. Sam Altman, OpenAI’s CEO, has been instrumental in this journey, taking OpenAI from a project that seemed uncertain of profitability to a valuation of approximately 80 billion dollars, a truly remarkable transformation. The journey of AI’s evolution continues to be filled with both intrigue and promise, and we look forward to sharing more of this remarkable story with you in our upcoming events.
2022 - 2023: The Year of AI
Let’s dive further into the rapidly evolving landscape of AI that has unfolded throughout the year. This year has undoubtedly been the year of AI, where the groundwork for transformative changes has been laid.
In January, Microsoft invested a staggering 10 billion in OpenAI, valuing it at a remarkable 40 billion. This marked a significant move in the tech industry, underlining the growing importance of AI. Microsoft didn’t stop there; they swiftly integrated AI tools into their existing product line-up, leading to announcements such as Teams Premium, setting the stage for the practical application of AI in everyday tools.
Google, recognizing the potential and fearing missing out, introduced Bard, although it couldn’t quite match the capabilities of Chat GPT, which also introduced a $20-a-month subscription model for early adopters.
Amazon was making moves in the background, hinting at their AI endeavours, while Hugging Face primarily served as a portal for developers seeking large language models, which proved invaluable for technical users.
Facebook, now Meta, ventured into the AI arena with Llama, initially launching it as open source, albeit accidentally due to a code leak. Their subsequent release of Llama Two, openly sourced, was a strategic move that seems to have worked in their favour.
Moving further into the AI landscape, Chat GPT and Whisper APIs took centre stage, facilitating text transcription and enabling AI to interact with other AI applications. Google offered Bard for free, signalling a commitment to democratise AI access. For enterprises, the launch of GPT for enterprise represents a significant step towards enhancing privacy and security in the corporate world. It’s crucial to understand that all these developments significantly shaped the AI landscape prior to November 2022, setting the stage for what was to come.
Before we delve into the post-November 2022 era, it’s essential to understand the AI landscape that many tech workers and knowledge workers were navigating. For tech workers, AI primarily meant retrieval AI, akin to Google search, with generative AI becoming accessible since 2017.
Generative AI, in essence, assists in coding and development, acting as a coding co-pilot, a tool that aids in the creation of code. Knowledge workers, on the other hand, relied on these AI-powered applications to consume information and perform tasks. However, post-November 2022 marks a shift in the AI landscape. Generative AI is democratising access to application development, making it easier and more cost-effective for anyone with basic computer skills to create applications, fundamentally transforming the way we interact with technology and opening new doors to innovation.
Giving Power to the People
The AI landscape is ushering in a paradigm shift, where natural language is becoming the new programming language. Instead of grappling with intricate programming constructs, we can now simply express our desire for an application’s functionality, and if done correctly, that’s precisely what we get.
This shift empowers us to swiftly and easily create personalised applications tailored to our individual use. Rather than relying on one-size-fits-all solutions, we can craft precise, specialised tools that align with our roles and tasks. This transformation carries two pivotal implications.
First, it signifies a dramatic reduction in the cost of software development. As early as 2023, we’re on the cusp of this shift, largely attributed to products like GitHub’s Copilot, which has been aiding coders since 2017 and enhancing their productivity by up to 40%. However, the real game-changer lies in Chat GPT and similar AI models, which will democratise the creation of applications for knowledge workers across various domains.
What was once an expensive endeavour, costing thousands, even hundreds of thousands. Will soon become an affordable, accessible endeavour, with each individual able to develop their specific programs for their unique needs
The second implication revolves around a substantial boost in knowledge worker productivity. A McKinsey study reveals that industries like Office and Administrative Support stand to experience a 45% increase in productivity, which resonates with many of us in this room. The report foresees AI automation potentially exposing 300 million jobs, but it’s important to note that this isn’t about job losses. Instead, it’s about unlocking the productivity equivalent of 300 million additional jobs. An OECD report even suggests a figure closer to 400 million, and this is within OECD nations alone.
The debate around AI’s impact on employment continues, but historical precedent suggests that while some jobs may be displaced, new opportunities and roles will emerge, ensuring a dynamic and evolving employment landscape. These are indeed exciting times, and as we navigate this uncharted territory, we are reminded to watch this space closely, as it unfolds with boundless possibilities.
Understanding Generative AI & Retrieval AI
Now, as we proceed, it’s imperative to grasp a crucial distinction that will shape our future interactions with AI. This distinction lies in the difference between generative AI and retrieval AI, and understanding this divergence is pivotal in navigating the AI landscape.
Up until this point, we have been predominantly using retrieval AI. When you perform a Google search or utilise Google Maps, you are engaging with a retrieval algorithm. It scours its data source, be it the internet or map data, to find relevant information.
For instance, if you search for “apple,” the algorithm looks for web pages mentioning apples and presents them to you in a ranked and sorted fashion. This is the fundamental operation of retrieval AI.
Google introduced what can be referred to as “semantic search.” Semantic search goes a step further by considering the context of the search term. For instance, if you search for “apple,” it might also suggest articles on fruits in general.
On the other hand, generative AI is a different beast altogether. It is trained on vast amounts of data, often involving terabytes of information from the internet. The key distinction lies in how this data is incorporated. In generative AI, the data is embedded within the model but not as a direct replication. Instead, it’s more like a massive spreadsheet, asking, “Given this set of words, what’s the next most likely word?” This approach is statistical in nature, as it is based on probabilities and predictions.
The beauty of generative AI is its non deterministic nature. You can ask it the same question multiple times, and it will provide slightly different responses. It’s unpredictable, yet it draws from the vast data it was trained on to generate its output. Think of all the training data as compressed into what we call “weights,” and these weights are used to generate further outputs. It’s a complex process that is challenging to condense into a single explanation, but the core point is that generative AI does not reference external data. Instead, it predicts what comes next based on the patterns it learned during training.
Conclusion
In conclusion, if you take away one thing from this discussion, it should be the distinction between generative and retrieval AI. Armed with this knowledge, along with AI literacy and a willingness to experiment, you can harness the transformative power of AI to enhance productivity, streamline processes, and drive innovation.
Feel free to scan the QR code in the room to access the slides and learn more about AI literacy courses. Your AI journey awaits, filled with opportunities to augment your capabilities and push the boundaries of what’s possible.
Following this event our Business Support will be hosting a bi-weekly event called ChatGPT 101: A Hands-On Experience to Boost Your Business
Want to find out more? Get in touch or come and join us at one of our upcoming events.
