961.Digital

Google has announced the launch of its chatbot, Bard, in response to OpenAI’s ChatGPT.

CEO Sundar Pichai wrote in a blog post that Bard is available to “trusted testers” and aims to provide access to the world’s knowledge through a conversational interface.

It utilizes a smaller version of the AI model LaMDA, similar to the technology used in ChatGPT.

However, Pichai did not mention integrating Bard into the search box, instead showcasing its use to enhance conventional search.

Google also plans to make the underlying technology accessible to developers through an API, but no timeline has been provided.

The chatbot technology is still prone to errors and cannot easily be updated with new information, presenting challenges for building powerful new products.

The popularity of ChatGPT has sparked speculation about Google’s dominance in web search being threatened for the first time in years. Microsoft, which recently invested $10 billion in OpenAI, is holding a media event today believed to be related to new features for its search engine, Bing, and its partnership with ChatGPT’s creator. OpenAI’s CEO, Sam Altman, tweeted a photo with Microsoft CEO Satya Nadella after Google’s announcement.

Launched by OpenAI last November, ChatGPT has gained widespread attention for its ability to answer complex questions with apparent coherence and clarity. However, AI experts warn against its limitations, as the tool does not truly understand the information it provides and is prone to fabrication.

Google’s AI researchers were involved in the development of the technology behind ChatGPT, a fact referenced in CEO Sundar Pichai’s blog post. Pichai emphasized Google’s investments in AI since reorienting the company around the technology six years ago and mentioned the company’s AI research division and DeepMind, the UK-based AI startup acquired by Google in 2014.

ChatGPT is based on GPT, an AI model that was first developed by Google to predict the next text given a string of text. OpenAI has made waves by showing how training transformer models with vast amounts of data and boosting computational power can result in language or image generation systems. ChatGPT improves on GPT by allowing humans to provide feedback to an AI model, which then fine-tunes the output.

However, Google has been cautious in incorporating the technology behind LaMDA into its products. AI models trained on web-scraped text can also generate incorrect information and display racial and gender biases as well as spread hateful language.

Google’s draft research paper in 2020, which highlighted these limitations, caused some executives to become upset and led to the firing of two leading ethical AI researchers, Timnit Gebru and Margaret Mitchell. Some Google researchers who worked on the technology behind LaMDA also became frustrated with the company’s cautious approach and left to start their own companies using the same technology. The emergence of ChatGPT seems to have motivated Google to accelerate its plan to incorporate text generation technology into its products.