This week, Bard—a Google competitor to ChatGPT—received an upgrade featuring a new feature called Bard Extensions. This feature allows the AI chatbot to connect with a user’s Gmail, Google Docs, and Google Drive accounts. Google has also granted Bard the ability to search YouTube, Google Maps, and other services, as well as introduced a fact-checking tool for Bard’s responses. The focus of this article, however, will be on the Gmail, Docs, and Drive integrations because the ability to ask an AI chatbot questions about personal data is incredibly valuable.
Bard Extensions addresses a common complaint with AI chatbots, namely that they exist in isolation. They lack the capability to view a user’s calendar, read their email inbox, or access their online shopping history—information that an AI assistant would need to provide optimal assistance. Google, with its vast amounts of user data, is well-positioned to bridge this gap. Many users, including myself, have their digital lives on Google’s apps and could greatly benefit from AI tools that leverage this data.
I put the upgraded Bard to the test and was met with mixed results. It performed well in simpler tasks, such as email summarization, but fell short in analytical tasks and provided incorrect information. Jack Krawczyk, director of Bard at Google, explained that Bard Extensions is primarily focused on retrieving and summarizing information rather than analyzing it. He acknowledged that the system might struggle with more challenging prompts, emphasizing that trial and error is necessary at this stage.
Currently, Bard Extensions is only available for personal Google accounts. Users must manually enable Extensions in the app’s Settings menu, and the feature is currently only supported in English. It’s important to note that Google has stated users’ personal data will not be used to train Bard’s AI model or be visible to employees reviewing Bard’s responses. However, Google advises against sending any data to Bard that you wouldn’t want a reviewer or Google to see.
While it’s safe to assume that most chatbots, including Bard, will improve over time, there are still privacy concerns that need addressing. Bard Extensions might not be a finished product yet, as indicated by the “Experiment” label and the disclaimer surrounding its limitations. Despite this, it’s surprising that Google released Bard Extensions in its current state, given its erratic behavior during testing. It seems the dream of a fully knowledgeable AI assistant that can flawlessly analyze our pasts and anticipate our needs is still a ways off.
Here are some tasks I attempted with Bard:
Task 1: Psychoanalyze Me
Using Bard Extensions, I asked Bard to analyze my Gmail and identify my biggest psychological issues. While Bard was able to generate a response, it seemed to be making things up rather than accurately analyzing my emails. When asked about a specific quote, Bard provided a fabricated sentence and email context. Mr. Krawczyk acknowledged that Bard’s performance might not be perfect and that improvements are still in progress.
Task 2: Be My Travel Agent
Bard now has access to Google’s travel products, including Google Hotels and Google Flights. In a demo video, Google showcased Bard as a useful travel assistant that can search through emails and find information about planned trips. During my testing, Bard produced mixed results. It correctly retrieved my flight dates but got the departing airport wrong. It also recommended a train that didn’t exist. Mr. Krawczyk suggested that this could be due to limitations with Google’s travel-booking apps, which currently don’t optimize around train schedules in Europe.
Task 3: Wrangle My Email
Given my struggle with email organization, I hoped Bard could assist me in decluttering and organizing my inbox. Bard performed well on simple tasks like summarizing recent emails from my mom or focusing on single subjects. However, when asked to perform more complex tasks, it stumbled. For example, when instructed to summarize the 20 most important emails, it included seemingly random emails. Similarly, when asked to draft responses to specific emails, it pulled from the wrong tab and composed a note to Nespresso instead. Generating a list of my most-emailed contacts also proved challenging for Bard.
In conclusion, while Bard Extensions shows promise, there is still work to be done to improve its accuracy and functionality. It’s clear that Bard is an experimental product, and users should approach it with that understanding. With time, it’s likely that Bard and other chatbots will improve, and privacy concerns will be addressed. The vision of a truly all-knowing AI assistant may still be a distant reality.
Denial of responsibility! Vigour Times is an automatic aggregator of Global media. In each content, the hyperlink to the primary source is specified. All trademarks belong to their rightful owners, and all materials to their authors. For any complaint, please reach us at – [email protected]. We will take necessary action within 24 hours.
