POV: AI Assisted Search

Background:

Generative AI has been the hot topic this week (again), particularly AI search. Both Microsoft and Google announced new features they say will revolutionise web search as the race for language-based AI dominance kicks off.

Details and Implications:

Microsoft declared its AI search intentions earlier this month, with the news that it planned to incorporate a faster version of OpenAI’s Chat GPT, known as GPT4, into Bing to give it more conversational and contextual results following its $10 billion investment into OpenAI.

Then, on Tuesday this week, the company held a press event for the big reveal of its revamped Bing search engine and Edge browser with built-in AI technologies. Key new features include AI-summarised search results, an interactive chat where you can refine your search and a ‘creative spark’ that suggests AI-generated ideas for your query. At the moment, users only get a few sample searches, and then they are prompted to join the waitlist.

Some of OpenAI’s GPT-3.5 language model was integrated into Bing. However, Bing’s version is more up-to-date and can handle queries related to more recent events, so it can provide information like pricing data or recent data for travel tips and itineraries, whereas OpenAI’s ChatGPT bot has so far only been trained on data up to 2021. Bing also cites its sources and links them in a ‘learn more’ section.

Microsoft CEO Satya Nadella acknowledged that, as with every new technology, it is important to remain conscious of the potentially negative consequences. He said: “It’s about being clear-eyed about the unintended consequences of any new technology.” He also  stressed that Microsoft wants to use technology that enhances human productivity and that is aligned with human values.

On Monday, Google announced its rival service ‘Bard’ and then gave a very brief demo of Bard at a YouTube live stream event in Paris on Wednesday. Like ChatGPT, Bard will query information from the web to "provide fresh, high-quality responses."  It will be powered by Google’s Language Model for Dialogue Applications (or LaMDA for short). The system is currently only being tested in closed beta, with wider public availability promised sometime “in the coming weeks”.

Along with the announcement of Bard, Google plans to add more AI-powered features to Google search, summarising information within the search page and providing deeper insights into the users' queries.

However, Google’s event fell a bit flat, with reports it was a bit underwhelming compared to the big show that Microsoft had put on the day before and things didn’t improve when an error was spotted in the promo video for Bard posted on Twitter. It wrongly claimed that the James Webb telescope was the first satellite to take pictures of a planet outside of the Earth’s solar system. This mistake was seen as an indicator of the underlying quality of the AI and saw stock in Google’s parent Alphabet drop by 9%.

Summary:

While concerns remain about the accuracy of AI’s conversational capabilities, Google and Microsoft are implementing AI to better understand and provide conversational search queries with more tailored and contextual responses. It’s likely to be just the beginning of a frenzy of rapid innovation in this space and how we as consumers discover and access information. It’s too early to draw any conclusions over where this latest arms race will take us or who will win, but for this week at least, it’s advantage Bing.

Further Reading:

Forbes | Adweek| TechCrunch | The Verge | Mashable