Microsoft and Google Give Search an AI Boost

Artificial intelligence (AI) is underpinning the future of Internet search, with Google and Microsoft announcing AI capabilities to their search engines this week. As noted in a previous blog post, AI has received substantial investment from cloud platform operators over several years, with Microsoft primarily investing in OpenAI and Google developing AI capabilities internally, as well as making smaller investments in start-ups such as Anthropic. In this article, we unpack Microsoft Prometheus and Google Bard, looking ahead to the future of AI and its place in the workplace.

Microsoft Prometheus

Prometheus, the model behind the new Bing experience announced on Tuesday, is powered by the latest OpenAI large language model, combined with frequently refreshed search data. Microsoft has applied the model to the core search experience, enabling “the largest jump in relevance in two decades” to ensure the accuracy and relevancy of search results. However, we note that the announcement focussed on the chat and content-generation capabilities.

Using the new chat feature in Bing, users can pose complex questions. Microsoft demonstrated potential uses like planning detailed trip itineraries and researching what TV to buy, showing the ability to refine the search until the user is happy with the answer. Users can also summarize contents of web pages, with earnings reports used as an example in the demonstration on Tuesday. Bing now offers functions similar to ChatGPT, such as composing paragraphs, e-mails or blog posts based on user prompts.

There are, of course, limitations to what it can do. The new Bing experience is quick to remind users that it’s a search engine, not a virtual assistant like Alexa, Cortana or Siri. It even refused to do an impression of 1980s fake-AI television personality Max Headroom (pictured above, as rendered by Stable Diffusion).

Google Bard

Google announced Bard, a conversational AI service, on Monday. The announcement came in a blog post by CEO Sundar Pichai that detailed the company’s history of AI-related research and productization. Mr Pichai noted that Bard will initially use a lighter version of Google’s large language model that needs less computing power and scales easily to more users.

Google also held an event in Paris on Wednesday, smaller in scope than Microsoft’s, where it announced relatively iterative upgrades to its existing products. The function attracting the most attention is “multisearch”, an AI-powered visual search system. It allows users to take a photo of what they want to search for alongside a text query, providing a modest upgrade to the capabilities of Google Lens.

The planned live demonstration of this at the event was abruptly skipped as the demo phone was missing. Also, a promotional video for Bard showed an incorrect answer, claiming that the James Webb Space Telescope was the first to photograph a planet outside our solar system, when in fact, the European Southern Observatory’s Very Large Telescope did this in 2004.

AI the Right Way

The incorrect answer publicly broadcast by Google Bard highlights concern about accuracy, or lack thereof. Information generated by AI has often been unreliable — it will be essential for users to independently review the accuracy of generated information.

AI ethics and regulations are becoming increasingly crucial as the technology becomes more widely used and integrated in society and business. Important considerations such as accountability and transparency, bias and discrimination, misinformation, security and privacy need to be addressed. Microsoft learned this in 2016 with its AI chat bot Tay, which was deactivated 16 hours after its launch for generating racist and offensive content.

In this rapidly evolving market, the real challenge will be the speed of these developments. It’s essential that the likes of Microsoft and Google put robust safeguards in place to ensure the technology is used in an ethical and responsible way. For example, Microsoft’s Office of Responsible AI was founded in early 2019 and works with OpenAI to engineer the necessary measures against harmful and abusive content. For its part, Google published a set of principles in 2018 to guide the responsible development and productization of AI.

Unlocking the Potential for the Workplace

Google and Microsoft appear to be focussed on the consumer market, but we shouldn’t forget that enterprises are looking for ways to stay ahead in the market and accelerate innovation. These recent advances in AI search tools have potential, and we expect they will play an important role in the future of the workplace, helping to improve productivity and streamline work processes.

Organizations are increasingly looking for ways to do more with less, and AI technology provides a solution to help employees find the information they need, saving time and increasing efficiency. However, it could be some time before AI search tools are embedded into businesses at a deep level — especially considering the risks of inaccuracies and the challenges of ethics and regulation such as intellectual property rights.

This technology is still in its infancy, and we can only imagine where it might take us in five, 10 or 20 years. We can be sure that it’ll change the way we research, develop and structure content in the future, facilitating differentiation and commercial opportunities.