Back in late 2022, the overnight success of ChatGPT energized Google to launch a chatbot just four months later. Gemini, named Bard at the time, has had a tumultuous journey since then, undergoing countless upgrades and an entire rebrand.
Also: OpenAI's new Model Spec reveals more about how it wants AI to behave
Despite all the large language model (LLM) upgrades, from LaMDA to PaLM 2 to Gemini Pro, Google's chatbot has failed to achieve the popularity of its rivals. With the company's annual developer event, Google I/O, just a weekend away, we can expect Google to roll out Gemini updates to make the chatbot more appealing to the public.
After tracking the evolution of Gemini since its launch and testing many other generative AI chatbots, I've rounded up some of the features Google could bring to Gemini to significantly improve the user experience, making the tool a more worthy chatbot competitor, and a potential replacement for my current AI default, Copilot.
The first and most obvious win would be for Google to make Gemini more enticing for iOS users. Since its rebrand in February, Google has been positioning Gemini as an AI assistant capable of going beyond what an ordinary chatbot can do -- and with Android phones, it has done just that.
Also: These 3 Gemini upgrades will help you get more out of the AI assistant
By downloading the Gemini app, Android users can take advantage of some neat integrations, such as accessing Gemini wherever they would regularly use their Google Assistant, activating an overlay experience that tells Gemini what is on their screen, and even accessing Google Assistant voice features, such as setting a timer on their phone.
iOS users, however, can't download a dedicated Gemini app; access to the chatbot is limited to the Google app. This is a huge missed opportunity given that many Apple users have Google as their default search engine. They could benefit from experimenting with Gemini as an assistant, especially without an Apple AI chatbot native to the iOS experience.
Microsoft Copilot is a good example of what Google could be doing with Gemini. The Copilot app currently ranks#22 in the App Store's Productivity category. If Google wants to market Gemini as an everyday AI assistant, iOS users should get a standalone app.
If you ask Gemini a question, the chatbot will answer without footnotes or source links. So that you can verify the accuracy and validity of its answers, Gemini offers a handy "double-check with Google" feature. The feature shows users where parts of the answers are sourced from. This extra step, however, requires an extra refresh, which can interrupt the user's workflow.
Also: ChatGPT vs. Microsoft Copilot vs. Gemini: Which is the best AI chatbot?
Furthermore, when you click the "double-check with Google" button, Gemini doesn't list all the sources. For parts of the response, the chatbot might say that Google Search didn't find relevant content (see the screenshot below). This response undermines the reliability of the answers and adds an extra layer of doubt when using the chatbot.
One of Gemini's advantages is that, unlike ChatGPT, it is connected to the internet. Google should capitalize on this functionality and create features to build trust with its audience, including clickable footnotes to source content without an extra step, a feature that Copilot already offers.
Gemini is already multimodal and supports the input of voice and image prompts in addition to text. As helpful as those two features are, the ability to import documents could help take the chatbot to the next level. Adding this feature would unlock a new suite of possibilities.
Also: 5 ways AI can help you study for finals - for free
Anthropic's Claude, for example, lets users import documents for free. This capability is one of the chatbot's biggest advantages because it allows users to work with materials they interact with daily.
Whether you want a research paper summarized, a wordy contract explained, or have questions about a PDF you're using, AI chatbots with document-reading capabilities can help.
Adding this feature would also give Gemini a competitive advantage over its biggest rival, ChatGPT, which only offers document uploading in the premium version of its chatbot, ChatGPT Plus, which costs$20 per month.
Users receive an ominous message when they open Gemini: "Your conversations are processed by human reviewers to improve the technologies powering Gemini Apps. Don't enter anything you wouldn't want reviewed or used."
Also: Gen Z looks to AI to overcome its data anxiety - but the rest of us are worried
When you click to learn more about the message, Gemini says: "If you don't want future conversations reviewed or used to improve machine learning models, turn off Gemini Apps Activity."
Yet when you go to that new window, you receive a contradicting message: "Your chats are saved in your account for up to 72 hours, whether Gemini Apps Activity is on or off. Google uses this data to provide the service, maintain its safety and security, and process any feedback you choose to provide."
So, it seems like Google will save your chats and continue to review them until the 72-hour mark. While it is true that generative AI models get smarter by learning from user inputs, many AI chatbots allow users to opt out of that feature entirely, having no chats saved at all.
Also: 4 things Claude AI can do that ChatGPT can't
For example, since April 2023, OpenAI has let users opt out of having ChatGPT use their data to train its models or save chats. Just this week, OpenAI updated ChatGPT's data controls even further by adding a Temporary Chat option for users who would prefer not to have their chats saved to their chat history even if the model improvement is turned off.
Generative AI users are typically more aware of their data privacy because they don't want their information used in future answers or shared with others. To encourage more use of its chatbot, Google should address privacy concerns and add a clearer and all-encompassing opt-out option.
As I mentioned at the beginning of this article, Google has been trying long and hard to popularize its AI models. As a result, the tech giant has implemented some of its generative AI offerings in its most popular product, its search engine, through the Search Generative Experience (SGE).
With SGE, users get an AI-generated answer to their search engine prompt at the top of search results. This is meant to provide quick, helpful, conversational answers that require less scrolling. However, public feedback suggests the experience is confusing and aggravating. Users find that SGE disrupts the regular search flow.
Also: Google was right to be worried: OpenAI reportedly wants to enter the search market
When Google first announced SGE, it was accessible through Google's Search Labs, where users would have to opt in to use the feature. Since then, however, many users have reported seeing SGE appear in their search results even if they hadn't opted in.
In March 2024, Google confirmed via a statement to Search Engine Land that a "subset of queries, on a small percentage of search traffic in the US" would get SGE. This forced exposure is leaving users with a negative opinion of Google's AI tech.
The solution here is simple. Google should focus on making its generative AI offerings, like Gemini, more enticing to users, so they want to join in, rather than forcing its AI features into a popular product, like Google Search.