AI has a new name: Super Intelligence. Now what?

On 29 September, the White House announced that artificial intelligence was over. Not the technology, which continues to do quite nicely, but the name.

President Trump signed an executive order directing federal agencies to call it ‘Super Intelligence’, or ‘SI’, instead. Apparently ‘artificial’ makes it sound fake, while ‘super’ is, in the president's words, "the best word of all".

HR and internal communications teams have spent two years persuading employees that AI isn't magic, isn't necessarily coming for their jobs, and still occasionally makes things up. Now there’s a new job: explaining why the same technology is officially super.

It is not only Washington. Silicon Valley is falling into line. Salesforce's Marc Benioff has renamed the company's AIForce agent platform SIForce, and Elon Musk has renamed SpaceXAI as SpaceXSI, its second rebrand in three months.

The question isn't whether "SI" will catch on. It's whether organizations need to respond, and how to do so without confusing employees, complicating compliance or undoing the work of building trust.

The most important decision may be what not to change.

Names are agreements, not labels

A name is less a word than an agreement between people about what they are talking about. The psycholinguists Susan Brennan and Herbert Clark called these "conceptual pacts". Once people have settled on a way of describing something, they keep using it, because it's the easiest way to be sure the other person knows what they mean.

‘AI’ is one of the most successful conceptual pacts of the decade. People who couldn't tell you what a large language model is will happily tell you they asked AI to plan their holiday. That's exactly when a name becomes hardest to shift.

Say ‘SI’ to most people and you'll have to follow it with "you know, AI".Say ‘SI’ to most people and you'll have to follow it with "you know, AI".

That doesn't make established names permanent, as a 2018 experiment published in Science showed. However, names change through repeated use, not by announcement.

And even when a new name does take hold, the old one lingers. Microsoft renamed Yammer as Viva Engage in 2023, and three years later employees I interview routinely call it Yammer. Twitter became X the same year, and I refuse to update the app on my phone because I prefer the little blue bird.

You can change the name on the building. Getting people to change the name in their heads is another matter.

When naming becomes political

We already know how a politically directed rename plays out. When the Gulf of Mexico became the Gulf of America in 2025, federal agencies switched, as did some platforms with commercial reasons to accommodate Washington.

The Associated Press declined and found itself barred from the Oval Office and Air Force One. Much of the rest of the world carried on as before. The new name didn't simply replace the old one. It became a tell: your choice said something about where you stood.

The SI rename is more complicated, because the US government is an enormous technology customer; suppliers have an incentive to speak their customers' language.

Once vendors adopt SI in their sales literature, it will travel into procurement, training and policy at organizations that never chose it. The administration doesn’t need everyone to say SI, only enough influential people that everyone else has to understand it.

Decide what actually needs to change

The first decision for HR leaders is whether anything needs to change at all.

US federal agencies must adopt the new terminology in materials covered by the order. Federal contractors and vendors chasing government business may find it commercially sensible to follow.

Meanwhile, anyone operating in Europe works under the EU AI Act, which defines an "AI system" in law and won't be redrafted to suit Washington.

That leaves global employers navigating two vocabularies. An organization might have an AI policy for its European workforce, SI terminology in documentation for a US government customer, and product-specific names in the applications employees actually use.

But different terminology doesn't automatically require a company-wide rebrand. Check with Legal and Procurement whether there is a genuine statutory, contractual or commercial requirement. For now, the order defines SI by reference to the existing legal definition of AI, so only the name has changed, though a new federal definition is due by late November.

The second decision is where inconsistency might actually matter to employees.

An employee completes AI training, encounters SI in a procurement document and opens Copilot to do their work. Are these three different things, or one? Does the acceptable-use policy cover all of them? The answers may be obvious to the people who wrote the documents, but not to those expected to follow them.

HR, IT and Comms should identify the points where conflicting terminology could cause real confusion: onboarding, acceptable-use policies, training, guidance, vendor documentation and support materials. Often the fix is a single sentence in an existing FAQ explaining that SI and AI mean the same thing, and making sure the helpdesk knows how to answer the question.

The third decision is who gets to decide what the organization calls it.

Without an agreed convention, Legal will use the language of regulation, Procurement will mirror suppliers, IT will follow product documentation and HR will stick with last year's training. Each function will be consistent with itself, while the employee experience becomes a muddle.

Agree a canonical term for employee-facing communications, define any necessary equivalents, and make someone responsible for maintaining consistency across guidance. This isn’t a branding exercise; it’s agreeing on a few words.

And there is a fourth option, one that organizations are often reluctant to consider: do nothing.

If there is no legal obligation, no material supplier change and no evidence that employees are confused, leave the terminology alone. Sometimes the most valuable contribution internal communications can make is to refuse to manufacture another change programme.

The trouble with ‘super’

The acronym is the smaller problem. The bigger one is the adjective.

The adoption work of the past two years has been an exercise in making AI boring. Not a revolution, not a threat, just a tool that drafts your email, summarises the meeting you missed and finds the policy you couldn't. Demystifying it is how you get past both the hype and the fear – and it's slow, unglamorous work.

‘Super’ pushes the other way on both fronts. Plus, ‘super intelligence’ already means something. In the debate about AI risk, it describes systems that outperform humans across virtually every cognitive domain, the thing people worry might outsmart us all. That's not the image you want in the heads of employees already wondering whether the new tool is sizing them up for replacement.

There's a more fundamental problem here than hype: successful adoption depends on people developing a reasonably accurate understanding of what a technology can and cannot do. They need to know when an output is reliable, when it needs checking, and when the tool is the wrong choice altogether.

So make trust calibration an explicit part of the adoption plan. Don't just measure whether people are using the tools; find out whether they know when an output needs checking, what information is safe to enter, and who remains accountable for the result.

If they don't, another awareness campaign won't fix it. Practical examples and clear guidance about where human judgement is still required will.

Name the tool, not the tech

The good news is that organizations have more control here than it might seem, because nobody actually uses ‘AI’ at work. They use Copilot, or Gemini, or whatever the internal assistant has been called.

The category is something people talk about. The tool is something they use.

In Digital Communications at Work, Jonathan Phillips and I argue that naming a channel is valuable because it gives people a handle for it. A name helps people picture what something is for and talk about it with each other. The same logic applies here. A clear, specific, stable name for the thing employees actually open every day does far more for adoption than any category label, and it is a name the organization can establish and reinforce.

So anchor your guidance, training and success stories in the tool rather than the technology. Write "use Copilot to summarise the thread", not "use AI to boost productivity". Explain what the tool does, where it fits into someone's work, and what they need to check before trusting the result. Those instructions remain useful whether the category is called AI, SI or something else entirely.

Resist rebranding your own programme: every rename restarts the conversation about whether this is something new.

The job of HR and Comms isn't to make employees fluent in the vocabulary of whoever happens to be in power. It's to help them understand and use the technology in front of them.

Call it Copilot, call it Gemini, call it whatever you like. Just make sure people know what it does, what it doesn't, and when they should trust it.

That's a more useful definition of intelligence.