The Gen AI wars have just taken a step up. To compete against the growing popularity of ChatGPT, (and of course Microsoft copilot which is powered by ChatGPT), Google announced that they are upgrading their Bard AI generative AI chat bot with the highly anticipated “Gemini” model, which Google say will create a vastly superior chatbot to OpenAI’s ChatGPT (in its current capacity at least).
As the dominant force in search, Gemini looks to help Google to better compete in the rapidly growing field of generative artificial intelligence and given their advertising reach, it won’t be long before everyone knows what Gemini is – kind of reminds me of the Genesys AI from the later Terminator Film 🙂
According to Google, the model has already made Bard smart enough to overtake OpenAI’s free ChatGPT service in six out of eight benchmarks it used including in math and language understanding.
In contrast, OpenAI’s ChatGPT v3 was released last year at end of November 2023 and has since been used as the primary large language model used by Microsoft Copilot. It has undergone many enhancements and upgrades this year including the most recent ChatGPT 4 turbo models.
ChatGPT can also be purchased and used as a standalone AI and comes in a free, Premium and enterprise SKU for business. Microsoft use ChatGPT4 in their consumer and enterprise grade generative chat services including Bing Chat (now Copilot for Edge) and Copilot for Microsoft 365.
What is Google Gemini?
Gemini is a collection of large language models that can handle various data tasks, as well as read and understand entire pages, including signature blocks, document stamps, process text, images, audio, video, 3D models, and graphs simultaneously.
Gemini comes in three tiers – Gemini Ultra, Gemini Pro, and Gemini Nano.
- Gemini Ultra is Google’s most powerful model, pitched as a competitor to OpenAI’s GPT-4.
- Gemini Pro is a mid-range model powered to beat out GPT-3.5, the baseline version of ChatGPT.
- Gemini Nano, a more efficient model built to run on mobile devices and is expected to come to Google pixel next year.
Google say that Gemini responds in a much more human like way to ChatGPT and has more sophisticated multimodal capabilities and can better master human-style conversations, language, and content.
The also say it can understand and interpret images, code prolifically and effectively, and generate data analytics. Gemini can also be used by developers to create new AI apps and APIs⁴.
Gemini is expected to power most of Google’s products and services in the near future, such as Google Workspace, Gmail, Search, Bard, Pixel, and Nest⁴. Some of the services will be included in their products for free whist others will be premium/costed options. It will be part of the Bard Advanced service. I couldn’t find any details around pricing at time of writing.
This is the biggest upgrade to Bard since it launched….. In the coming months, users can also expect Gemini to power other services, including Google Search, ads, and Chrome.
Google
According to Google, the Gemini model has already made Bard smart enough to overtake OpenAI’s free ChatGPT service in six out of eight benchmarks it used including in math and language understanding. They claim:
“Gemini is the first model to outperform human experts on MMLU (Massive Multitask Language Understanding), one of the most popular methods to test the knowledge and problem-solving abilities of AI models.”
Demos and teasers
People can see demos of Gemini’s capabilities on a website that showcases how their new AI model can understand user inputs in a variety of different sources, including text, images, audio, and code. For example, the model can produce a small computer program simulating the movement of a flock of birds simply by showing Gemini a video.
A YouTube video also shows a teaser aimed at parents in which it shows how Gemini can understand math problems written on a sheet of paper and provide correct answers.
Gemini – a threat to Microsoft ChatGPT and Copilot?
In the consumer world, where Google dominates search, this could stop Microsoft from making any form of dent in Google’s search dominance, and area Microsoft has been trying to gain market share in with the power and uniqueness of ChatGPT 4 through Copilot in Edge. ChatGPT already has a strong brand presence in the Enterprise and Commercial space and also leverages Bing as its primary internet search tool, making the relationship between Open AI (the founder of ChatGPT) two-way and plutonic.
What is ChatGPT 4?
ChatGPT-4 is an AI-powered language model developed by OpenAI, capable of generating human-like text based on context and past conversations. At time of writing, it is the most advanced system produced by OpenAI, with broader general knowledge and problem-solving abilities than previous models. It can solve difficult problems with greater accuracy, generate, edit, and iterate with users on creative and technical writing tasks, such as composing songs, writing screenplays, or learning a user’s writing style. It is more creative and collaborative than ever before. It can also generate visual content and solve problems with visual input. GPT-4 is 82% less likely to respond to requests for disallowed content and 40% more likely to produce factual responses than GPT-3.5 on OpenAI’s internal evaluations.
ChatGPT-4 is built on the structure of GPT-4, which is based on a large language model that checks for the probability of what words might come next in sequence. It enables users to refine and steer a conversation towards a desired length, format, style, level of detail, and language.
What is Microsoft Copilot?
Microsoft Copilot is the name for Microsoft’s suite of AI tools built closely around the ChatGPT large language model. Initially announced in March 2023 and now generally available in other product suites include Edge, Power Platform and GitHub, Copilot is now a board room level conversation for most commercial organisations.
Copilot for Microsoft 365 for example brings generative AI into the heart of Microsoft’s collaboration and productivity tool set and combines that with the organisations content, data and objects such as people and relationships into the Microsoft 365 apps and services that people use every day, such as Word, Excel, PowerPoint, Outlook, Teams, and others.
Microsoft Copilot is also not just a single AI model but is described as a platform that also empowers organisations and third-party developers to you to build their own copilots for different domains and scenarios.
Microsoft also recently released Copilot Studio to create custom copilots that can handle specific tasks, such as sales, cybersecurity, healthcare, education, and more⁸. There is then the wider Azure AI Studio which can be used to build generative AI apps, discrete and trainable language models and custom copilots that using natural language prompts.
Summary
As you can see, Google Gemini, ChatGPT and Microsoft Copilot have their own strengths and weaknesses, and they can complement each other in different ways. The only way to determine which is best for each business use case (and it may not be a simple case of one or the other) is to look at the use case, test and PoC each service (and others) and make your own deductions. Cost and integration, as well as security, control, and governance will also be key considerations as organisations look to develop and adapt their strategy around AI and in particular Gen AI tools within business.
Happy AI-ing!
Links, References and further reading
Microsoft Copilot for Microsoft 365 overview | Microsoft Learn.
Introducing Gemini: Google’s most capable AI model yet (blog.google)
Chat GPT Enterprise by Open AI