It’s hardly new that the business world sometimes takes a while to embrace the newest and most innovative tools. One tool that stands out for its potential revolutionary impact, yet currently faces inertia, confusion, and even some resistance, is Copilot in Excel.
In this post, I aim to distinguish Copilot for Excel from Microsoft’s array of other Copilot tools, unveil the origin story of Copilot, delve into some prevalent doubts about Copilot for Excel, and share my perspective on why this product is still a wise one to invest in.
Who is Copilot for Excel intended for? What is its purpose?
Copilot for Excel is designed for a wide array of users, including professionals, students, and anyone eager to enhance their spreadsheet management. It assists with common tasks such as sorting, filtering, and formatting data, as well as more advanced functions like creating detailed analyses and visualizations. This tool streamlines workflow and improves productivity, accommodating both routine and complex spreadsheet operations.
Since Copilot is designed to enable users to engage with the software through natural language and conversation rather than through traditional coding, menus, or computer-centric methods, it is theoretically suitable for users of all skill levels. However, the ability to formulate natural language in a manner that computers can effectively interpret is an important factor to consider.
From Clippy to Copilot
Though Copilot represents a groundbreaking advancement in technology, the concept of a natural language-driven assistant for Excel has been anticipated for quite some time. A notable example of an earlier effort is Clippy, the animated paperclip assistant introduced by Microsoft in the late 1990s.
Clippy was designed to provide help and tips via a conversational interface, aiming to make software more accessible and user-friendly. Despite its well-meaning objectives, Clippy often fell short of providing genuine assistance. Clippy’s failure to resonate with consumers was primarily due to its habit of offering unsolicited and irrelevant advice, a consequence of its inadequate natural language processing capabilities. It was unable to learn from interactions or tailor its assistance, leading to repetitive and unhelpful suggestions.
Moreover, its attempt to seem friendly and approachable with an anthropomorphic design ended up being more annoying than beneficial. The combination of its lack of sophistication in understanding user needs, its intrusive presence, and the difficulty in disabling it rendered Clippy more of an obstacle than an aid for many users.
The advent of OpenAI’s GPT algorithm marked a revolutionary advancement in natural language understanding and generation. This generative AI model is capable of comprehending and producing human-like text based on the input it receives. The distinction comes from the architecture and the vast scale of data on which GPT was trained, enabling it to grasp context, infer meaning, and generate relevant and coherent responses. This represents a significant leap over predecessors like Clippy in terms of capability and flexibility.
Copilot, a contemporary tool designed to operate with Excel (among other applications) using natural language processing powered by technologies akin to GPT, signifies the progress made in aligning computing with human thought processes.
Unlike Clippy, Copilot doesn’t merely wait for users to pose questions or offer unwelcome advice. It can understand complex queries, execute tasks based on natural language instructions, and even generate insights and analyses from data. This is feasible because generative AI, the foundation of Copilot, comprehends language nuances, processes vast amounts of information, and learns from interactions to provide increasingly pertinent and accurate assistance.
The transition from Clippy to Copilot highlights the evolution in user interface design from basic attempts at comprehending human language to sophisticated interactions facilitated by generative AI. This shift is underpinned by advancements in machine learning models capable of processing and generating language in ways previously unimaginable. Consequently, tools like Copilot are not merely more potent versions of their antecedents; they embody a fundamentally different approach to human-computer interaction, one that is more natural, intuitive, and efficient.
The Copilot product line
Microsoft first introduced the term “Copilot” with GitHub Copilot, a tool launched in June 2021. GitHub Copilot is an AI-powered code completion tool designed to help developers write code more efficiently by suggesting entire lines or blocks of code as they type. It’s based technologies developed by OpenAI and can generate code in a variety of programming languages and styles, making it a versatile assistant for coding tasks.
Since the introduction of GitHub Copilot, Microsoft has expanded the “Copilot” branding to other products, notably within its Microsoft 365 suite. Some of these products include:
- Copilot in Windows
- Copilot for Microsoft 365
- Copilot for Sales
- Copilot for Service
- Copilot for Finance
- Copilot in Viva
I think it’s fantastic that Microsoft is so optimistic about the future of AI, investing heavily in it. However, the marketing approach of naming everything so similarly despite differences in access, licensing, and pricing models, is quite confusing.
To clarify, Excel is part of the Copilot for Microsoft 365 suite, along with other familiar desktop productivity tools such as Word, Outlook, and PowerPoint. This means you cannot gain access to Copilot in Excel as a standalone offering; it is packaged together with the other Microsoft 365 Copilot tools. However, it is important to note that this is a completely different product from GitHub Copilot or even Windows Copilot.
How to get Copilot 365
Access to Copilot in Excel is available through Copilot for Microsoft 365. To purchase Copilot for Microsoft 365, a product license for Microsoft 365 Business Standard, Business Premium, E3, E5, Office 365 E3, or E5 is necessary. The cost is $30 per month. To add Copilot to your existing enterprise subscription, please contact your Microsoft representative. For more information about pricing, licensing, and getting started, check out the 365 for Windows page.
LLMs and user adoption of Copilot for Excel
Responses to Copilot, including both Copilot for 365 and Copilot for Excel, have been varied. Regrettably, Excel has emerged as one of the less favored tools, receiving considerable criticism from users across a broad spectrum of expertise:
I must admit, my initial encounter with Copilot brimmed with anticipation. Envisioning it as a tool poised to revolutionize the way I automate, organize, and debug my workbook, I expected a level of sophistication akin to ChatGPT’s capabilities.
However, Copilot’s debut revealed its proficiency in only the fundamentals—sorting, filtering, conditional formatting, and crafting basic charts for individual tables. Beyond these basics, it struggles, marked by its sluggishness and the unpredictability of its outputs.
This performance, understandably, might stir frustration. The slow response times, inconsistencies, and confined utility of Copilot in Excel invite comparisons to Clippy.
But this isn’t a fair comparison. At its core, Copilot is fueled by generative AI, a technology celebrated for its adaptability, learning capacity, and the promise of progressive enhancement. This foundational difference suggests a potential for Copilot that Clippy could never aspire to—a future of refined functionality and responsiveness, sculpted by user feedback and iterative learning from vast data.
To dismiss Copilot now is to overlook the essence of generative AI’s journey—an evolution from a nascent stage to maturity, much like a seedling’s growth into a tree. It’s reasonable to feel let down by Copilot’s present limitations, especially when expectations aren’t met. However, recognizing the trajectory of AI development offers a broader perspective.
Interacting with Copilot, despite its flaws, fuels its evolution. Each piece of feedback and every interaction is a step towards refining its algorithms, boosting its efficiency, and customizing its utility to better align with user demands.
Large Language Models (LLMs) serve as complex, data-trained systems capable of understanding and generating human-like text.. Every question answered, every mistake corrected through user interaction, enriches the LLM’s grasp of language, making it increasingly nuanced and adept.
The future of Copilot for Excel
I remain highly optimistic about Copilot and Excel, despite recognizing that there’s a considerable journey ahead. Currently, it might not meet everyone’s expectations or needs, but it’s essential to acknowledge the ongoing development. Even Microsoft has openly admitted the need for further progress.
During the Morgan Stanley Technology, Media, and Telecom conference in March 2024, Jared Spataro, the corporate vice president of Modern Work & Business Applications at Microsoft, admitted that Copilot in Excel needs to evolve:
So if you look at Copilot in Excel, like, we all can’t wait until it’s an Excel jockey and can do a lot, but it’s coming. It’s learning and people have very high expectations. Same is true of PowerPoint and it’s going to disappoint there because we’re still learning the commanding surface of Excel.
The acknowledgment of these challenges, as popular wisdom holds, is the first step toward improvement.
Microsoft’s investment in AI technology is strategic and reflects the rapid evolution of this field. The excitement around AI and its inevitable integration into Microsoft’s future developments is undeniable. This is particularly apparent with the Office suite’s trajectory, which is increasingly focused on Copilot.
I envision a transition towards an AI-centric experience with Copilot, leading to a seamless integration of AI functionalities in Excel. This integration will transform Excel into an inherently AI-powered tool, distinguished by its sophisticated capabilities.
In the future, Excel might develop the ability to create intricate and visually appealing charts, (perhaps through Python or other programming langauges), formulate complex equations, and even generate complex pro forma financial forecasts, reports and more by understanding the input data’s structure and company goals.
While achieving these milestones may seem ambitious given Copilot’s current state in Excel, laying a solid foundation is crucial for rapid future development. Microsoft is committed to enhancing and supporting Copilot for the long haul.
With all that in mind, it’s still a smart move to get started early. It’s beneficial to become adept at interacting with Excel using natural language, and to develop a strong intuition by repeatedly iterating through a series of prompts. This approach will help you learn how to structure your data optimally for the best outcomes when working with AI.
What are your thoughts on LLMs, Copilot, and the future of Excel? Do you foresee a promising future, or are you skeptical about its potential? How do you envision Excel’s evolution? Share your views in the comments.
Leave a Reply