Wednesday, February 08, 2023

How will Google and Microsoft AI chatbots affect us and how we work?


Dan Milmo Global technology editor, and Kari Paul in San Francisco
Tue, 7 February 2023 

Google and Microsoft are going head to head over the future of search by embracing the technology behind artificial intelligence chatbots.

Google announced on Monday that it is testing Bard, a rival to the Microsoft-backed ChatGPT, which has swiftly become a sensation, and will roll it out to the public in the coming weeks.

And on Tuesday, Microsoft announced it is increasing its focus on artificial intelligence, boosting funding for new tools and integrating the technology underpinning ChatGPT into products including its Bing search engine and Edge browser, with the goal of making search more conversational.

ChatGPT, developed by San Francisco company OpenAI, has reached 100 million users since its public launch in November, becoming by some estimates the fasting growing consumer app of all time.

Here are some questions about Google and Microsoft’s AI plans and their likely impact.

Why are Google and Microsoft using AI in search?

The reaction to ChatGPT shows that there is an appetite for AI-enhanced search and for answers to queries that are more than just a link to a website. Microsoft clearly sees this as a competitive opportunity, as does Google judging by its rapid response. Google also believes users increasingly want to access information in a more natural, intuitive way (using tools such as Google Lens, which allows people to search using images and text).

Dan Ives, an analyst at the US financial services firm Wedbush Securities, says: “While Bing today only has roughly 9% of the search market, further integrating this unique ChatGPT tool and algorithms into the Microsoft search platform could result in major share shifts away from Google.”

What is the technology behind the Google and ChatGPT chatbots?

Bard and ChatGPT are both based on so-called large language models. Google’s is called LaMDA, an acronym for “language model for dialogue applications”. These are types of neural networks, which mimic the underlying architecture of the brain in computer form. They are fed vast amounts of text from the internet in a process that teaches them how to generate responses to text-based prompts. This enables ChatGPT to produce credible-sounding responses to queries about composing couplets, writing job applications or, in probably the biggest panic it has created so far, academic work.
How will Bard be different from ChatGPT?

Google has yet to make Bard publicly available but it uses up-to-date information from the internet and has reportedly been able to answer questions about 12,000 layoffs announced by Google’s parent, Alphabet, last month. ChatGPT’s dataset – in the form of billions of words – goes up to 2021, but the chatbot is still in its research preview phase.

Google’s chief executive, Sundar Pichai, said Bard could answer a query about how to explain new discoveries made by Nasa’s James Webb space telescope to a nine-year-old. It can also tell users about the best strikers in football “right now” while supplying training drills to emulate top players. The screenshots supplied by Google showed a more polished interface than ChatGPT’s, but it is still not accessible to the public so direct comparisons with the rival OpenAI service are difficult.

How will the technology behind Bard and ChatGPT change Google and Microsoft’s search engines?

Google says its search engine will use its latest AI technologies, such as LaMDA, PaLM, image generator Imagen and music creator MusicLM. The example presented by Pichai on Monday was a conversational, chatbot-like response to a question about whether it is easier to learn the guitar or the piano. It appeared at the top of the search query instead of, for instance, a link to a blogpost or a website. Again, Google has not released this AI-powered search model to the public so questions remain.

Microsoft detailed its revamp of Bing on Tuesday, announcing that it will be able to answer questions using online sources in a conversational style, like ChatGPT does now. It will also provide AI-powered annotations for additional context and sources, perhaps reflecting concerns among some ChatGPT users about the accuracy of some user answers.

“It’s a new day in search,” said Microsoft’s CEO, Satya Nadella, at an event announcing the products. “The race starts today, and we’re going to move and move fast.”

Will generative AI transform our jobs?


Generative AI, or artificial intelligence that can create novel content ranging from text to audio and images via user prompts, is already having an impact, and has stoked fears it could replace a range of jobs. BuzzFeed will use OpenAI technology to enhance its quizzes and personalise some content, according to a memo obtained by the Wall Street Journal.

BuzzFeed’s chief executive, Jonah Peretti, said humans would provide ideas and “cultural currency” as part of any AI-powered creative process. In Hollywood, AI is being used to de-age actors while ITV has created a sketch show based on deepfake representations of celebrities.

Michael Wooldridge, a professor of computer science at the University of Oxford, said some industries were going to feel a significant impact.

“Generative AI will have big implications in some industries – those who write boilerplate copy for a living are going to feel the influence soon,” he said. “In web search, it will make browsers much better at understanding what we are searching for and presenting the results in a way we can understand – just as if we asked our query of a person, rather than a machine.”

He added that ChatGPT and other similar systems have flaws and can get things wrong, as users of the OpenAI chatbot have found.

“Treating them as sages is really not a good idea,” he says. “Until we know how to make them reliable, this is not a good use of the technology: best stick to the things it is really good at, like summarising a text and extracting key points from it.”

No comments: