Artificial intelligence: will it ever be able to write emails for us?

Artificial intelligence: will it ever be able to write emails for us?

Artificial intelligence

What if your inbox is flooded with artificial intelligence (Ai) generated emails? It is possible that you have already received written messages from the AI, albeit with the help of a human prompter. Austin Distel, marketing director at Jasper , is one such person.

While flaunting Jasper's email composition prowess, Austin smiles: “These are tools that have helped me work faster, but also better,” he says, adding that he often uses AI to rewrite his work emails to match the tone of US comedian Jerry Seinfeld.

The role of autocomplete

Emails are already one of the least genuine forms of communication today. The phrases of circumstance and the courtesy answers with which we pepper them have a robotic and involuntarily comic quality (why, for example, do people keep writing things like I hope this email finds you well?). In recent years, Google has integrated machine learning into its email service, Gmail, to generate one-sentence responses and predict what users are likely to type. More recent companies like Compose Ai may soon allow people to rely more heavily on autocomplete features when writing emails.

“Autocomplete won't take you away from what you already want to type. It will just speed it up,” explains Michael Shuffett, founder and CEO of Compose Ai. Although large blocks of text generated with AI from tools like ChatGpt are very popular, at the moment the short sentences created with autocomplete are easier to control. The difference is the same as going on a tandem with a robot and leaving it alone on your bike at night after giving it a one-sentence command. The first option requires more effort, but increases the chances of reaching the intended destination together.

Major email providers, such as Gmail and Outlook, will introduce new AI-powered features to help users unravel their overflowing mailbox? Aparna Pappu, vice president and general manager of Google Workspace, points to several ways the company is using AI to help users compose messages and detect spam. In a statement sent to US (obviously via email), it emerges how his attitude towards generative AI reflects most of the messages coming from Google on the subject: enthusiasm, but also caution.

“We are in a new era of AI, where large language models have the power to take utility to the next level,” Pappu writes. “We know it is essential that this work is done with the utmost attention to safety, quality and validity". Companies using Microsoft's customer relationship management system, Viva Sales, can test a generative AI tool for writing response emails. Among the possible pre-packaged requests to be submitted to the AI ​​are discount offers and answers to questions. Microsoft declined to submit a comment for this article.

Not So Near Future

The future of email, therefore, may feature more comprehensive autocomplete. But would it be a revolution? Not exactly .

It would be if a custom AI model proved capable of composing high-quality, comprehensive emails by mimicking your writing style. Office workers who spend their days typing answers on their laptops would effectively have a digital secretary at their disposal.

Several startups are moving towards this reality. In 2019, well before the advent of ChatGpt, serial entrepreneur Matt Shumer was sitting in front of his computer responding to a slew of emails sent from associates, clients and investors. The activity seemed a bit overwhelming to him, so when OpenAi made its Gpt-2 model public, Shumer wondered if it could be used as a productivity tool.

He then retrieved several pieces of information from his mailbox and used them to train a model. "Then I asked him to write new emails – he says -, and I was immediately amazed by what I saw". This experience later led him to become the co-founder and CEO of OthersideAi, the company that developed HyperWrite, a generative AI tool capable of composing entire paragraphs and rewriting complicated sentences.

Despite the prospects of a future populated by automatic email authoring software, the majority of marketing professionals and CEOs interviewed for this article admitted that the technology has not yet reached a level sufficient to be able to entrust them with complete control of an inbox email. Philippe Lehoux, CEO of the Missive work messaging company, declares: " This technology will improve, but anyone who aims to use AI to replace themselves and give automatic replies right now is deluding themselves "; John Humphrey, product manager of email marketing platform Mailchimp , likens generative AI to an intern who happens to come up with brilliant ideas, but who you would never trust to manage a major project.

So what what does the technology excel at? In addition to writing responses, two other emerging use cases for email-related AI are changing the tone and creating potential subject lines. Sure, generative AI can reword your emails to look like they were written by Jerry Seinfeld. But with an alternative tip, he's also able to draft messages that are more direct, serious, or apologetic. But always be sure to re-read the post before hitting “reply all” as large language models continue to have a tendency to make up facts.

AI tools can be used to analyze the data and propose a hundred ideas for the subject of an email. “The principles behind what makes an object effective? AI hasn't changed them. It's just made it easier to produce them,” Humphrey points out. It will then be up to marketers to decide which angle is most suitable for their clientele.

Finally, company executives should keep in mind the controversies related to texts generated by artificial intelligence, to avoid alienating the recipients of their emails. In the US, Vanderbilt University has apologized after using ChatGpt to message students about the recent Michigan State University shooting. People who approach generative AI as a tool to find the right words to talk about dramatic events may misunderstand the strengths of the technology, which is able to predict the words to insert in a sentence and speak in general about the human experience , very useful for trying to better understand potential buyers, but not so much for trying to connect with people after a tragedy.

This article originally appeared on US.

Powered by Blogger.