While talking to people, I found out that many people know about chatGPT, but not many people know about Bard.
There are many who think that Google is completely behind in the AI race...
It seems like there was some prejudice due to the impact of LaMDA or PaLM or typos during the keynote.
Personally, I think Google's Bard is better for use in Korea than chatGPT . *As of June 11, 2023
Officially, "Korean" is managed separately from token management.
Provides multiple types of answers rather than just one answer
Partially integrated with Google real-time search
Used in Gmail, etc. by linking with Google Workspace, etc.
ChatGPT is great, but Bard is... useful a lot (from a general user's perspective)
If you can use APIs or plug-ins well, the GPT-4 model is obviously effective.
Many users don't know anything about LLM, let alone coding, so Bard has a lot of immediate uses in terms of usability.
Of course, MS is partially introducing Copilot, and there are expectations that it will get better once it is fully implemented.
In particular, Bard's upgrades in this update include improved abilities in math, coding problems, and string manipulation.Added abilityto export to Google Sheets
A new technique called 'implicit code execution' gives Bard the ability to detect computational prompts and execute code in the background, improving the accuracy of its responses.
Large language models (LLMs) act as prediction engines and have shown high performance on language and creative tasks, but have been relatively weak in the areas of reasoning and mathematics. (In fact, in many cases, they cannot even solve simple addition problems in mathematics.)
Bard improves reasoning and mathematical abilities through a novel way of generating and executing code, an approach based on Daniel Kahneman's theory of the separation of 'System 1' and 'System 2' thinking from Thinking, Fast and Slow .
Bard combines the power of LLM (System 1) with the power of traditional code (System 2) to improve the accuracy of responses, which improves the accuracy of responses to word and math problems based on calculations by about 30% (still showing hallucinations - symptoms of lying)
๐ Welcome to Haebom's archives. --- I post articles related to IT ๐ป, economy ๐ฐ, and humanities ๐ญ. If you are curious about my thoughts, perspectives or interests, please subscribe. Would you like to be notified when new articles are posted? ๐ Yes, that means subscribe. haebom@kakao.com
