What's your take on Google Gemini's potential?
Sort By:
Oldest
C-PIO in Software9 months ago
Google is trying to get a share of the AI market. Microsoft has established itself as the leader with open AI. Hopefully this version of googles AI will be better than before.Associate Director of Software Engineering in Finance (non-banking)9 months ago
I’m pretty impressed with the video they posted, although it is hard to evaluate it just based on the video I am certain that it pushes forward the competition. Chief Technology Officer in Finance (non-banking)8 months ago
Google’s Gemini is a highly capable AI model that has been built from the ground up for multimodality, meaning it can reason seamlessly across text, images, video, audio, and code. It is Google’s most advanced AI model and has outperformed human experts on MMLU (Massive Multitask Language Understanding), a popular method to test the knowledge and problem-solving abilities of AI models.Gemini surpasses performance on a range of benchmarks including text and coding. It has shown promising results in various fields such as general knowledge representation, reasoning, reading comprehension, commonsense reasoning for everyday tasks, basic arithmetic manipulations, challenging math problems, and Python code generation.
The potential applications of Gemini are vast. It could supercharge Google’s AI business and there might be advanced developments in the field of medical science, as medical AI chatbots and robots could potentially run on Gemini. However, the full potential of Gemini will only be realized as it is further developed and applied in various fields.
The baseline, Gemini is one of the many tools where you should evaluate according to your use case at that point in time (since tech is evolving too fast, this comment may be obsolete in months if not weeks/days)
Chief Information Officer (and Branch Manager)4 months ago
We did a 90 day pilot of Gemini as part of a test case and to inform a Business Case we are building. Having completed the 90-day pilot, I would say we saw mixed results.
I will begin by qualifying things to say, as we are in Canada, the ongoing (but now settled) standoff between our Federal Government and some key providers (Google included) that was not related to AI still had a significant impact on us. Google as a corporate entity chose to delay releasing features to Canada until that was solved.
As such, our test period had some users test the version Canada was granted before the Federal item was settled and some tested the version after.
The version afterwards, meaning the generally commercially available model delivered adequately.
Challenges we saw and encountered:
1—It never worked, as shown in the videos. Videos show a lot of integration into Google Workspace products that, simply put, isn't quite there yet.
2—As is 'The Google Way' during the pilot period, features would come (and occasionally go), and accessing them would get changed. Sometimes, it was a button you'd hit; other times, to get access to the features inside another product from Google (e.g., Sheets), you had to dig into menus and sub-menus.
3 - Related to #2, the story most saw about the early challenges with Google and things like novel image generation, resulting in Google temporarily deprecating that ability was annoying.
Now that said, I think these are generally growing pain type challenges, and, in part, I think because Google had to rush this to market perhaps a bit quicker than they'd have liked,and, also, because the evolution in this space is at a pace we've not seen in tech in a while.