Google Bard vs ChatGPT: Battle of the AI Chatbots – Medium
Sign in
Sign in
Member-only story
Gennadiy Shevtsov
Follow
—
Share
The field of artificial intelligence (AI) has witnessed a remarkable surge in innovation, with ChatGPT from OpenAI leading the charge. However, Google entered the scene with Bard AI, launching in March 2023. This article explores the key distinctions between Google Bard and ChatGPT based on recent testing conducted in July 2023.
Initially powered by Google’s LaMDA and PaLM 2 language models, Bard has evolved, now utilizing Gemini Pro for generating responses. In our tests, Bard demonstrated a slight edge in providing more nuanced and detailed responses compared to ChatGPT. The use of diverse data sources, including science papers, maths expressions, and source code, gives Bard a unique approach to constructing answers.
The fundamental difference lies in the training data sources. Bard is trained on data selected to enhance both dialogue and coding capabilities, supplemented by real-time internet access. Gemini Pro outperforms its predecessor, PaLM 2, particularly in coding queries and written tasks. In contrast, ChatGPT, powered by GPT-3.5 and the exclusive GPT-4 for ChatGPT Plus users, relies on a predefined dataset, including Common Crawl, Wikipedia, news articles, and various documents.
Beyond the data sources and language models, there are additional differentiators between Bard and ChatGPT:
ChatGPT has a history of generating code, but Bard’s switch to PaLM 2 significantly improved its coding capabilities. Gemini Pro, according to early reports, further enhances Bard’s coding proficiency.
ChatGPT can learn and retain context from conversations, while Bard does this in a more limited way.
Bard allows users to access multiple responses to a query through “View Drafts,” while ChatGPT generates a single response.
—
—
My journey through the realms of programming has led me to establish a dynamic digital haven:https://omniscopelife.com/
Help
Status
About
Careers
Blog
Privacy
Terms
Text to speech
Teams
