Generative AI Face-off: Microsoft Copilot vs. ChatGPT vs. Google Bard

Introduce the term “generative AI” into casual conversation and expect a range of responses. For some, generative AI promises to revolutionize creativity and productivity in astonishing ways. Others fear the loss of jobs, the potential for increased cyber threats, or bias and invasion of privacy.

Generative AI will revolutionize productivity and innovation in many positive ways. It will also change some jobs, and threat actors already use it to manipulate and misinform. Additionally, like all emerging technologies, there are ethical considerations to consider. Regardless, the technology has arrived, and it has great potential for good when used wisely.

Three of the most well-known generative AI models include Microsoft Copilot, Google Bard, and ChatGPT. While all three interfaces remain under constant development, this post offers a brief introduction and comparison.

Feature-rich Microsoft Copilot Delivers Seamless Integration

In the past year, Microsoft has begun making good on its promise to deliver AI capabilities throughout its product suite. Consequently, it offers various flavors of generative AI, from the free version (also known as Bing Chat) to GitHub Copilot and the subscription-based Microsoft 365 Copilot.

Microsoft Copilot currently generates text, code, and images. It also serves as a useful research tool. And Microsoft recently announced a new plugin for Bing Chat that will generate music.

Strengths

• Wide access – Based on GPT-4, it includes access to the rest of the internet.
• Lists sources for research – In chat mode, Copilot cites sources for the information it presents, allowing users to verify the information’s authenticity.
• Ability to choose a conversational style – Users can choose among creative, balanced, or precise styles. Additionally, when composing text, they can specify the tone, format, and length of the text to create. For instance, a user might ask Copilot to write a short, professional email or generate a funny list of ideas.
• Code generation – Copilot does an excellent job of creating code snippets in multiple languages based on natural language prompts. It also offers real-time code completion and suggestions, generating code based on context and the user’s style.
• Integrates smoothly with Microsoft 365 – Microsoft 365 Copilot works seamlessly within Microsoft applications. It includes the ability to pull data from across the Microsoft 365 suite. For example, it could create a list of pros and cons for a specific topic based on information in recent emails, meeting minutes, and chats.

Limitations

• Accuracy – Copilot’s suggestions, whether for research or for code generation, may include inaccuracies. Users should always review content and code before adopting it.
• Many features require a subscription – While anyone can use basic Copilot/Bing Chat for free, additional features require a subscription.

Google Bard Gaining Speed

While Copilot hit the ground running, Bard lagged behind for a while. However, recent developments show it catching up and developing some impressive capabilities. Users currently access Bard for free primarily through the Chrome browser, although Google is exploring wider availability.

Bard currently generates text, code, and images. It also serves as a useful research tool. And Google recently launched a beta of MusicLM that allows music generation.

Strengths

• Informative – Bard does a good job of providing clear, informative answers, though it does not currently cite sources. It presents information in an easy-to-understand format, such as creating a table to compare features of similar products.
• Understands complex queries – Trained on a massive dataset of text and code, Bard can grasp nuanced questions and provide comprehensive answers.
• Ability to pull data from popular apps – Bard responses can now include real-time data from Gmail, Google Maps, Flights, Hotels, and YouTube. It can also summarize content in Google Docs and Google Drive. For instance, a user might ask Bard to find and summarize the most recent email from their supervisor.

Limitations

• Limited feature set – Compared to paid versions of Copilot, Bard offers limited functionality, though it compares well to the free version. For instance, Bard still lags behind Copilot in generating code.
• Does not cite sources – Unlike Copilot, Bard does not include source citations. It does offer a double-check feature, but that feature has been known to make mistakes.
• Questionable accuracy – Although Bard says it strives for factual accuracy, it often delivers inaccurate responses that sound quite convincing.
• Currently only available through Google Chrome

ChatGPT: Engaging and Creative

An early leader in the generative AI space, ChatGPT continues to shine as a conversational agent and a creative assistant. Users can access both a free version and a subscription version either through the OpenAI website or through a mobile app on iOS or Android.

ChatGPT currently generates text and images. While not designed for coding, it can generate snippets of code in certain circumstances.

Strengths

• Engaging conversationalist – Users like ChatGPT’s ability to carry on natural-sounding conversations, adapting tone and style to the context.
• Content creation – ChatGPT does a good job at generating creative text.

Limitations

• Questionable accuracy – ChatGPT emphasizes fluency over facts, so users may prefer using Copilot or Bard for research purposes.
• Limited feature set – ChatGPT lacks several key features and integrations offered by Microsoft Copilot and Google Bard.

Enhance Business Operations with Generative AI…But Use Caution

Businesses currently use generative AI to drive efficiency and innovation. For instance, airports use AI to enhance identity verification systems. Customer service chatbots offer personalized assistance. And generative AI works well for creating realistic audio with text-to-speech generation.

However, organizations should do their homework and learn the risks before fully embracing AI. For instance, the content generated through AI may prove inaccurate or biased, potentially contributing to data compliance issues. And the use of AI opens yet another doorway for threat actors to exploit.

Use the information in this post to determine which generative AI engine works best for you. And working with technology experts like eMazzanti enables organizations to determine the most productive use cases. Tap their expertise to implement effective cyber security and information governance controls, to reduce the risks involved.

Download Article PDF