Google may be stumbling out of the gate a bit with its latest large language model chat bot Bard, but it looks like the company may already be looking for a successor with an all-encompassing feature set in Project Gemini.
According to Luke Jones over at WinBuzzer, and Jon Victor over at the Information, Google is quietly working on a much more encompassing artificial intelligence platform dubbed Gemini which the company hopes can compete with the likes of Bing Chat and ChatGPT.
Specifically, the team from Google’s DeepMind division is currently in a bunker crafting a “natural language processing system that can generate natural and engaging responses to user queries, as well as provide relevant information from various sources.”
While Gemini might sound like Google’s current Bard effort, it actually looks to be the evolution of the chatbot industry geared toward the multitude of input scenarios that are growing with uses.
Gemini is intended to go beyond predefined scripts, templates, and generally trained data sets, and offer more personalized and relevant answers to queries such as travel information geared toward a specific users’ preferences rather than a summation of traveler’s recommendations.
Google is training Gemini using a myriad of data sets but a standout one is the use of YouTube video transcripts that help the platform provide advice based on video content.
Furthermore, with access to Google’s substantial internet archives and search engine platform, Gemini can also provide additional context to queries with images, videos, blogs, and relevant app recommendations.
However, the biggest leap forward for Gemini comes in communication recognition which will allow it to understand and respond with similar accents, regional dialects, slang, idioms, abbreviations or even emojis.
Anecdotally, my wife tends to use emojis to accent her texts while I use GIFs, and it seems Gemini could learn from similar communication trends when accessed and respond in kind, along with replicating accents from verbal interactions.
There isn’t a clear plan or roadmap for Gemini just yet, but Jones believes the Gemini will be used across many of Google’s products such as Bard, email assistant, and search, which all make sense.
According to Jones and Victor, Google is sprinting to get Gemini up and running and in front of users quickly and possibly as soon as this Fall with a budget in the range of tens or hundreds of millions of dollars behind it.
As an aside, Google was among the first companies to craft large language models but held off for years implementing its efforts due to privacy, intellectual property and misinformation concerns as well as its potential to cannibalize its vast search engine revenues.
Google’s temped Bard offering may be the company dipping its toes in the water of AI chatbots, but it looks like Gemini make be its cannonball dive into the market, but it remains to be seen how the company navigates its traditional search engine business with the omnipresence of AI-backed info at users’ fingertips and voice activated devices.