Google’s new Gemini update to their so-called “ChatGPT killer” Bard claims to make it more powerful than its competitor, but says so in corporate-speak that makes little narrative sense, and in language that sounds like it was written by AI.
We are now in the phase of the artificial intelligence (AI) boom where the big players like Google, Microsoft, and Salesforce are announcing products whose descriptions don’t make much sense in plain English, but investors are going crazy throwing money at them. As the AP explains in their coverage of Google’s latest AI product Gemini, Google’s market value has increased this year by around 45%, or “more than $500 billion," because of investor interest in their AI products, despite that those products are getting their collective asses whooped by ChatGPT and its next generation, GPT4. Meanwhile Microsoft, a major investor in ChatGPT creator OpenAI has seen its market value grow by 50% in 2023.
Gemini 🤝 Bard
— Google (@Google) December 6, 2023
Starting today, our specifically tuned version of Gemini Pro is available in Bard, unlocking new ways to collaborate with AI. Next year, we’re introducing Bard Advanced with Gemini Ultra for even more complex tasks. Learn more ↓ #GeminiAI https://t.co/hEPbj9faHr
Google released its ChatGPT competitor Bard in March, and it has not exactly taken the word by storm. Mindful of this, TechCrunch reports they just released their update called Gemini, which Google’s research claims is more powerful than ChatGPT.
After more than eight months of research and development, Google has unveiled its most powerful A.I. so far. But only parts of it are widely available. https://t.co/vk4ccKxNKD
— NYTimes Tech (@nytimestech) December 6, 2023
The New York Times has the technical specifics of Gemini (and the new name may be a rebranding effort of a middling product). The Times says Gemini os “available to English speakers in more than 170 territories and countries, including the United States, beginning immediately,” which is another way of saying it will only work in English at the outset.
The Times adds that Gemini will also be able to incorporate images and sounds instead of just text, “but those skills will not be rolled into the Bard chatbot until a later date.”
So has anyone actually tried Gemini yet?
— Nick Dobos (@NickADobos) December 6, 2023
-Pro only in bard
-Bard still not actually updated
-Only text capabilities, no multimodal
-no api yet
So NONE of the demo video are actually possible to try…
Google once again has a blog post for a waitlist for Gemini ultra???
So Google is largely selling features that have not been rolled out yet. And as TechCrunch notes, Gemini will be released in three tiers (Ultra, Pro and Nano), respectively referring to use in enterprise-level, desktops, and mobile devices, though Pro is the only one being released right now.
The AP report notes that Gemini could eventually be “infused into Google's dominant search engine,” which may or may not be good news to those who feel that Google’s search results have been getting worse over the last few years.
So I just asked Gemini Pro in Bard for the updates regarding the hottest news story these days, the war in Israel and Gaza, and its answer essentially was, I kid you not, "just Google it." 🤦♂️ pic.twitter.com/GZthPlOLzI
— Bojan Tunguz (@tunguz) December 6, 2023
The incorporation of images and sound may amplify the legal risk over swiping copyrighted work for profit, without compensating the artists and creators from whom the AI products swipe.
But the bigger issue with AI is that generative AIs so far have no idea when they are spouting complete bullshit. (Engineers prefer to call this phenomenon “hallucinations” rather than “complete bullshit.”) Fixing that problem would be the real breakthrough in AI, but there is little appetite to fix it among tech companies and the investor set thus far.
OpenAI's Sam Altman recently discussed the phenomenon of AI hallucinations on the Hard Fork podcast, saying that while chat bots may get some stuff wrong, the reason that ability is in the code is because sometimes you want the AI to get creative, and sometimes you don't.
Related: AI Engineers Convinced They Can Fix SF’s Problems at Hackathon [SFist]
Image via Google