Microsoft turns Copilot into a misanthrope's waking nightmare

The US software giant has unveiled a slew of new features for its Copilot AI solution that are designed to make human-computer interactions not just more productive, but to also make them feel as natural as possible.

Central to this endeavour is Copilot Voice, which gives users a choice of four AI voices with which to converse. Available on mobile, it can turn any simple query into a long-winded conversation, and could prove irksome to people who turn to electronic devices specifically to avoid the idiosyncrasies inherent to human conversation.

In the example served up by Microsoft, a woman is having a mini-crisis in a store. She needs to buy her non-wine-drinking friend a house-warming present, but – horror of horrors – wine is what she usually gives to people as gifts.

Lacking the initiative to ask the shop assistant, the woman instead asks Copilot Voice. It suggests alternatives like artisanal teas, food hampers or fancy olive oils – adding that Tuscan is its 'go-to' olive oil. Because if there's one thing we've all learned about AI, it's that it is very discerning when it comes to salad dressing.

"This is the most intuitive and natural way to brainstorm on the go, ask a quick question or even just vent at the end of a tough day," claims Microsoft, overlooking the centuries of social interaction that preceded generative AI (GenAI).

Copilot has also been added to Microsoft's Edge browser. By adding '@copilot' into the address bar, the AI can summarise the content of a Website, translate text or answer questions. Great news for the dozens of people who use Edge.

Taking things a step further is Copilot Vision. This is where it gets really weird. Copilot Vision can interpret whatever it is that is on-screen, and engage in a voice conversation about it.

In Microsoft's demo, the user discusses some old pictures of his grandmother, who Copilot describes as "a fashionable lady," and having "such a timeless style," before helping him out with deciphering one of her recipes for German pancakes.

Microsoft is treading very carefully with Copilot Vision, restricting what content can be analysed, and ensuring that any images are not used for training data. Conversations take place locally and are then deleted upon conclusion. At the moment it is also only available via Microsoft's new early access service, Copilot Labs, which lets eligible users play around with some of its more experimental features.

"I truly believe we can create a calmer, more helpful and supportive era of technology, quite unlike anything we've seen before. Great technology experiences are about how you feel, not what's under the hood. It should be about what you experience, not what we are building," said Mustafa Suleyman, EVP and CEO of Microsoft AI, in a blog post.

"Copilot will be there for you, in your corner, by your side and always strongly aligned with your interests. It understands the context of your life, while safeguarding your privacy, data and security, remembering the details that are most helpful in any situation."

Microsoft isn't stopping there. In addition to serving up to users a summary of relevant news and information – a feature called Copilot Daily – it is also rolling out Think Deeper. Also only open to Copilot Labs users, it claims to be able to reason through more complex questions.

With Think Deeper, Microsoft says Copilot can help with all manner of practical challenges, like "should I move to this city or that?" or "what type of car best suits my needs?"

Any responses to these queries that Copilot serves up will presumably be distilled from its vast archive of training data – which includes a lot of the same online resources someone would use were they conducting research the traditional way. Nonetheless, it would be fairly eyebrow-raising to base a major life decision on the recommendation of an AI with an affinity for olives from central Italy.

All these new features sound like they would be novel to use. For a while. But the jury's out on whether Copilot Voice and Vision will have greater staying power than earlier virtual assistants like Cortana, which never took off and was killed off when Microsoft launched Windows 11.

Microsoft recently launched a new fund that aims to raise $100 billion to help pay for the energy and compute resources that AI is expected to consume going forward. $100 billion to enable people to chat about house-warming gifts and pictures of grandma. It's hardly the most compelling argument for all that spending.