- The Alignment
- Posts
- Sundar Pichai: ‘We clearly have more capable models’
Sundar Pichai: ‘We clearly have more capable models’
PLUS Bloomberg's news FinanceGPT and Selfies from historical events
Happy Monday ! In today's issue we'll discuss whether Bloomberg is trying to make its own analysts obsolete and Sundar Pichai's confidence that they have the best models in the industry.
As always if you enjoy reading our posts be sure to spread the word !
Here’s what we have lined up for you today -
Sundar Pichai : ‘We clearly have more capable models’
Bloomberg releases BloombergGPT, financial analysts obsolete?
Selfie from the Last Supper
It’s hard to believe that Bard is the best that Google has to offer.
Sundar Pichai explained that the current version of Bard is run on a lightweight version of the LaMDA model. Upgrades will be coming soon and Bard should be getting better each day as Google deploys PaLM into Bard, its more recent and powerful model.
Pichai admitted that “In some ways I feel like we took a souped-up Civic and put it in a race with more powerful cars.”
Pichai has also said that Larry Page and Sergey Brin have been more actively involved in the company’s latest developments ever since the founders went into semi-retirement a few years back.
Pichai explained that Google’s reason for holding back its full AI capabilities has been because of its concerns with the societal impact these models can have. According to Pichai, AI is moving too fast and he is concerned that the pace poses a threat from bad actors who have access to these models. It must also be noted that because of Google’s size and market dominance, the company has been scrutinised heavily by governments all over the world and one wrong move by the company could set them back by a lot. ChatGPT (which is far less scrutinised) has already been banned in Italy !
Google Search and Workspace is the greatest corpus of training data there could ever be. Google’s also got the best AI minds working for them. But could the drawbacks of being a massive organisation (and the different levels of decision making that come with it) along with heavy government scrutiny become their achilles heal in this AI race ?
Bloomberg has developed a large language model specifically trained on a wide range of financial data. The model has been built to support various NLP tasks within the financial industry.
The model is aware of the specific terminology used throughout the financial industry. Not only is the model able to do all of the standard NLP tasks that Bloomberg was performing, it will also be able to comprehend the large corpus of information that the Bloomberg terminal handles, this is especially important for Bloomberg's customers.
Bloomberg largely developed the model in-house, and has been able to produce state of the art results on finance tasks. With over 4 decades of financial data, Bloomberg’s model will only continue to improve.

Bloomberg has been able to spin up a powerful model relatively fast. This model is able to perform financial tasks that are usually done by junior analysts. Are these analysts now going to lose their jobs, or will their productivity soar through the roof ?
Around the industry -
Stability AI CEO says AI will prove more disruptive than the pandemic
OpenAI geoblocks ChatGPT in Italy after ban from Italian Government
Stability AI CEO has the ambition to IPO in next few years
ChatDoctor - A medical chat model fine tuned on LLaMA
Product Corner -
There are way too many cool products to showcase just one !
YakGPT : Run ChatGPT locally on your browser and speak to it instead of typing
SplineAI : Design faster in 3D with prompts
Fyli : Build personalised chatbots without writing a line of code
Revive : Teach decision making with the help of AI
Shutterstock AI : Generate AI images on demand ready use to use immediately
Draftlab : AI copilot for Gmail, write and reply to emails 10x faster.
Selfie from the last supper
If selfies existed back in the day how would major historical events look like ?
Check out a few more here !
