Google has finally decided to answer the challenge posed by Microsoft-backed OpenAI and its ChatGPT AI chatbot. The search giant confirmed it will soon start public testing of its own chatbot, called Bard, which is based on the Google’s Language Model for Dialogue Application (LaMDA). Shruti Dhapola takes a look at the development
What is Bard?
Bard, Google’s own AI chatbot, is based on LaMDA. In a blog post, Alphabet CEO Sundar Pichai termed it an “experimental conversational AI service”, and said that Google will be “opening it up to trusted testers ahead of making it more widely available to the public in the coming weeks”. Given LaMDA has been in testing for nearly two years, the Bard rollout is quite fast in comparison.
As per the blog post, Bard “draws on information from the web to provide fresh, high-quality responses”. It will give in-depth, conversational and essay-style replies, just like ChatGPT. However, Google has said that the model is currently a “lightweight” version of LaMDA, and “requires significantly less computing power, enabling us to scale to more users, allowing for more feedback”.
Is Bard better than ChatGPT?
Running these models requires significant computing power. For instance, ChatGPT is powered by Microsoft’s Azure Cloud services. The service runs into errors at times, when too many people access it. Google says it will look at external feedback, along with “internal testing to make sure Bard’s responses meet a high bar for quality, safety and groundedness in real-world information”.
Google is looking for a lot of feedback at the moment on Bard, so it is hard to say whether it can answer questions better than ChatGPT. Google has also not made clear the amount of knowledge that Bard possesses. For instance, with ChatGPT, we know its knowledge is limited to events till 2021. Bard is also built on Transformer technology — the backbone of ChatGPT and other AI bots. Transformer technology was pioneered by Google and made open-source in 2017.
Google showcased several capabilities of LaMDA last year, including a new project called Wordcraft, used to help write fiction. Last September, Google revealed that it “teamed up with professional writers who used the Wordcraft editor to create a volume of short stories.”
These stories are available online for reading. But Google had also cautioned that LaMDA was not very good at writing fiction by itself and was more of a helper to human writers.
Why has Google announced Bard right now?
Google’s Bard announcement comes as Microsoft is preparing to integrate ChatGPT into its Bing search engine. Microsoft has invested $10 billion this year in OpenAI, and the ChatGPT-Bing integration poses a headache for Google and its Search business.
Google might have invented the ‘Transformer’ technology, but is now seen as a latecomer to the AI revolution. Many feel ChatGPT is “the end of Google Search”, given that it can give long, essay-style and sometimes elegant replies to a user’s queries. Of course, not all of these are correct, but AI is capable of correcting itself and learning from mistakes.
For Google, this has resulted in a ‘code red’, as The New York Times reported. In fact, another NYT report said that Google called in its founders Larry Page and Sergey Brin to review AI plans. The founders have mostly stayed off the day-to-day running of Google, but, clearly, some sort of alarm bells have gone off. Many perceive ChatGPT as superior right now, and the onus is now on Google to prove that LaMDA and Bard can do better. Microsoft also plans to offer ChatGPT to its enterprise customers as part of the Azure Cloud services.
Google has also announced it plans to bring AI features to search results.