Who still sends emails? In the unicorns of Silicon Valley, this practice is on the way out. Slack made it cheesy with its trendy business email. Start-ups no longer swear by it. And the big groups are now following suit. Acquired by Salesforce in 2020 for the tidy sum of 27.7 billion dollars, the company has revolutionized the way of working and communicating in business in a few years. Its cutting-edge teams are now working hard on the next work revolution: that brought about by generative artificial intelligence (AI) like ChatGPT, which Slack is already beginning to integrate into its platform. Interview with its co-founder and CTO Cal Henderson.
The Express: The Generative AIs keep progressing. Did they experience such a boom in 2023 surprise you?
Cal Henderson: What’s surprising is how they’ve captured people’s imaginations in six months and the impact they already have on a daily basis. OpenAI has been around for a long time. The progress they were making was very interesting, but so far it has been difficult to identify concrete applications for these advances. ChatGPT has changed the game. Giving these AIs a face, in the form of a chatreally showed what it was possible to do with it.
You recently conducted a study on how the French perceive AI. What comes out of it?
The surprise is that the French are more optimistic than the Americans about AI. 73% of executives in France say that artificial intelligence is the most effective technology for saving them time, and 63% say they are more productive thanks to it. A majority of French working people believe that AI is a major step forward for their country – 70% of executives and 62% of employees. The results were much more mixed in the study we conducted in the United States.
The latest advances in AI have aroused much criticism and fear. Which do you think are valid? And conversely, completely irrational?
Two questions arise in this area. The first is how the AI will progress. It’s very difficult to say! But the horizon of AGI [NDLR : artificial general intelligence, une intelligence artificielle générale capable d’apprendre des tâches variées à l’instar d’un humain] is probably further than my life expectancy! ChatGPT looks like magic but it isn’t. And it is very far from what is designated by the term AGI. ChatGPT will however bring many changes, including positive ones. I think he will create ultimately more jobs than it will eliminate. But some jobs will be disrupted. It is understandable that this type of rapid upheaval creates uncertainty and concern.
What should we urgently change in our educational approach?
Education will have to change and in particular the way students are assessed. For what ? Because generative AIs can easily fool our current evaluation methods. That’s not a bad thing. The ways children learn and assess change over time, as do the kinds of skills that are useful to teach them. The calculator or the Internet have already produced similar changes. Computers are very good at multiplying large numbers, so it is no longer relevant to spend as much time as before training people in this. We must look closely at the areas in which humans are the only ones to excel. On the other hand, it is a mistake to think that we would no longer need people with technical skills. The technique infuses in all trades, and there will be an increasing demand for these profiles.
What are the most interesting applications of generative AI in the professional context?
I see three of them: synthesis, composition and the search for information. Synthesis is the ability of these AIs to transform a long text or an exchange between many people into a summary of twenty, ten or even two lines. Humans are great at summarizing, but if we can automate these tasks, we’ll save valuable time. What I call composition is the ability to give these AIs a simple sentence and ask them to write a longer, personalized text, a formal email or a warm message for example. Of course, this may give rise to some grotesque situations. You will ask your AI to create a long email from a short sentence containing your main idea… and your recipient will open this email and ask their AI to summarize it in one sentence (laughs). But there will be great uses for it, especially for marketers looking to personalize the content sent to customers. Finally, generative AIs can be used to find relevant information in varied content sets. Using AI on your company’s Slack will identify the expert person on a specific subject, the status of a project or the actions taken by your company the last time it faced a certain type of issue. As a business historian you might question.
Slack has profoundly changed the way of communicating in business. Are you going to kill e-mail?
Email is not going away anytime soon, it will be around for at least fifty years. Look at the fax. It played a much smaller role than e-mail in business, yet it still exists. Email has been around for decades, it’s rooted in our communication habits. Even if it goes out of fashion, even if young people use it less, it will take several generations for e-mail to disappear completely. And it is very difficult to predict what the next communication trends will be. The new generations like to send voice notes to each other, to the great surprise of people from mine who are too reminiscent of the answering machine. Email is not just an old technology, it is also a model of the past. This is the digital version of the mail. This does not reflect the way people communicate in the office which is much more relaxed and collaborative.
What major changes has technology brought about in the way we work?
The desktop computer has changed the way we work for twenty-five years, but especially the way we communicate within the company. The pandemic has accelerated ongoing transitions to geographically decentralized teams and asynchronous rhythms by five to ten years. If the pandemic had hit even five years earlier, businesses would have had a hard time staying productive, the negative effects would have been much deeper. In 2020, the situation was very complex for sectors such as distribution. But in the sphere of information professions, productivity has almost improved, which was a real surprise. In recent months, it has of course been generative AIs that have been reshuffling the cards. They are further accelerating the movement towards the automation of certain tasks. Over the past twenty years, more and more functions have been augmented and partially automated by technology. Humans spend less and less time on repetitive tasks and more and more time on those where only they excel, such as creative and collaborative work.