B17
Registered User
- Posts
- 143
- Posts Power
- 143.0%
- Liked
- 1
- Joined
- Sep 14, 2023
- Bonus Points
- 7,125
Google is trialing its GPT-4 alternative, Gemini, among a small group of companies, according to a report from The Information. While the company is dominant in search, it has been caught off-guard with respect to generative AI and has been trying to respond to OpenAI and Microsoft’s partnership.
According to reports, Gemini is a collection of large language models (LLMs) that perform tasks people are already familiar with such as content creation, coding help, and more. Google has apparently given companies access to a large version of Gemini, but not the largest which is on par with GPT-4.
While Gemini is only being tested among businesses right now, it is expected to be available for all consumers in the near future. It’s expected to be integrated throughout Bard, Google Docs, and Slides. Over time, it could be baked into more products; a sensible candidate would be Gmail.
According to Android Authority, Gemini could be capable of more than just generating text. It could possibly create contextual images, analyze charts, create graphics with text descriptions, and control software with text or voice commands.
One of the issues around generative AI that has caused a stir in recent months is access to training data. One advantage that Google has is that it has YouTube. It has been using the transcripts from videos to train Gemini so that it can provide advice based on videos. Apparently, it’s avoiding copyrighted videos, though, so that it doesn’t get into hot water afterward.
While many search engines such as Bing and Brave Search have already started using AI features in their search results for all users, Google has been less keen to mess about with its search experience so users don’t feel alienated.
It has launched Search Labs in the United States where users can turn on experimental generative AI features in search and elsewhere. However, in most of the rest of the world, Google Search is still without any generative AI features. Hopefully, once Gemini is done, Google will speed up the rollout of some of these features.
According to reports, Gemini is a collection of large language models (LLMs) that perform tasks people are already familiar with such as content creation, coding help, and more. Google has apparently given companies access to a large version of Gemini, but not the largest which is on par with GPT-4.
While Gemini is only being tested among businesses right now, it is expected to be available for all consumers in the near future. It’s expected to be integrated throughout Bard, Google Docs, and Slides. Over time, it could be baked into more products; a sensible candidate would be Gmail.
According to Android Authority, Gemini could be capable of more than just generating text. It could possibly create contextual images, analyze charts, create graphics with text descriptions, and control software with text or voice commands.
One of the issues around generative AI that has caused a stir in recent months is access to training data. One advantage that Google has is that it has YouTube. It has been using the transcripts from videos to train Gemini so that it can provide advice based on videos. Apparently, it’s avoiding copyrighted videos, though, so that it doesn’t get into hot water afterward.
While many search engines such as Bing and Brave Search have already started using AI features in their search results for all users, Google has been less keen to mess about with its search experience so users don’t feel alienated.
It has launched Search Labs in the United States where users can turn on experimental generative AI features in search and elsewhere. However, in most of the rest of the world, Google Search is still without any generative AI features. Hopefully, once Gemini is done, Google will speed up the rollout of some of these features.