Yesterday, March 20, 2023, Google released BARD (Bayesian Actor Reparameterization Distribution). Like other assistive like Bing Chat and OpenAI (ChatGPT) BARD wants to be your personal assistant. You can plan your 10-year-old’s birthday party, provide themes and a supply list or write a personalized cover letter for a job you are applying to or generate a dinner recipe with the items you have in your refrigerator. (You can also do this on Bing Chat and ChatGPT). As an educator, I personally appreciate AI assistance for inspiration. I have used ChatGPT to create interactive lesson plans and group activities. BARD and other AI assistance claim you can even grade papers, if you have tried this, I would love to hear about your experience.
In an article on BBC, Google senior product director Jack Krawczyk said (BBC, 2023):
Bard is “an experiment” and he hopes people will use it as a “launchpad for creativity”.
Google says Bard is programmed not to respond to offensive questions and there are filters to stop it from sharing harmful, illegal, or personal information – but vice president of Google Research, Zoubin Ghahramani said like any methods of protection these could sometimes not work. (Para. 5)
BART is experiencing the same ethical concerns as Bing Chat and OpenAI but has the benefit of hindsight after watching others launch first.
How is BARD different than a Google Search?
BARD differs from Google search by the way it generates information. For example: If you want to write a poem, you can provide a prompt and it will write a personal poem for you instead of providing links to websites that have scripted poems. (You can also do this on Bing Chat and ChatGPT).
How is BARD different than other generative AI?
BARD has a “Google it” button which allows users to fact-check what sources have been used, such as Wikipedia. BARD also has an age requirement of 18 years of age. (Although, it didn’t ask me my age when I signed up for the waitlist.)
According to ChatGPT, BARD is different in approach and in the dataset it uses. Here is what Chat GPT (2023) says about those differences:
In terms of comparison to Chat GPT, the two models differ in their approaches to language processing. While Chat GPT uses a transformer-based architecture that allows it to model long-range dependencies in language, BARD utilizes a Bayesian framework that improves its ability to handle uncertainty and limited data. Additionally, Chat GPT is trained on a massive dataset that includes internet texts, books, and other sources, while BARD is trained on a smaller dataset that includes news articles, academic papers, and other sources.
Both models have their strengths and weaknesses, and the choice between them largely depends on the specific task at hand. For instance, Chat GPT may be more suitable for text generation tasks that require long-term coherence, while BARD may be better suited for tasks that require more robust handling of uncertainty and limited data. Ultimately, both models represent significant advancements in natural language processing and are valuable tools for researchers and businesses looking to leverage the power of AI to process and understand language data.
Want to try an AI Assistant?
Get on the waitlist for BARD
Sign Up for ChatGPT
Get on the waitlist for Bing Chat
This blog is part of a series on AI at: www.angelalcraig.com. Other blogs you might find helpful:
Unlocking Potential: The Benefits of Generative AI in Supporting Students with Learning Disabilities
Explore the Power of AI: A Directory of Websites and Applications Organized by Category
Maximizing the Potential and Minimizing the Risks of Generative AI: One Educator’s Perspective
References:
BBC Newsround (March 22, 2023). Accessed at: https://www.bbc.co.uk/newsround/65036003
ChatGPT (March 22, 2023). Accessed at: https://chat.openai.com/chat
Graphic:
Created by DALL-E AI using the prompt: Create a modern graphic with the word, BARD.
What do you think?