My two current obsessions have collided: artificial intelligence and watching what Donald Trump is doing next.
In an 8 October Truth Social post, Trump said the White House considers people who use “artificial intelligence” rather than “super intelligence” to be “THE ENEMY”. So, apparently, I am in trouble.
It is a memorable line. But there is a problem with the proposed name: superintelligence already means something in AI discussions.
AI, AGI and superintelligence
Artificial intelligence is the broad label for systems that perform tasks we associate with intelligence. The tools businesses use today sit somewhere under that broad umbrella.
Artificial general intelligence, or AGI, is usually the idea of a system with broadly human-level capability across many different tasks. Artificial superintelligence goes further: capability beyond humans across a wide range of tasks. Researchers do not all use exactly the same definitions, but the distinction is useful.
If we call every AI tool “super intelligence”, what do we call the genuinely more capable thing? Super-duper intelligence?
It is like deciding every car is now a supercar. Lovely: my car has had an upgrade without leaving the driveway. But what do we call a Ferrari? Changing the badge does not change what is under the bonnet.
What actually matters to a business
At Smart Cow we research new AI tools and test what they can actually do for marketing and running a business. The question is less exciting than a new name, but more useful: can the tool do worthwhile work accurately and consistently?
For example, could it help you:
- follow up a new enquiry quickly and in the right tone;
- turn your notes into a first draft of a quote;
- suggest and organise marketing ideas; or
- take a repetitive admin task off your plate?
Those are possibilities to test, not promises that every tool will get them right. A plausible-looking answer can still contain a mistake, miss context or invent a detail. Keep a person responsible for checking anything that goes to a customer.
A simple test you can run this week
Choose one job you repeat often. Write down how long it normally takes and what a good result looks like. Try an AI tool on that job, then check the output against your normal standard. Correct any mistakes and record the total time, including checking and fixing.
If the result is good and the full process saves time, you have found a useful application. If it needs so much correction that it takes longer, change the workflow or move on. That tells you far more than a label.
Watch the video for the full take, including the car analogy. If you want to discuss a practical AI use case in your business, talk to Smart Cow.
Sources: Trump’s original post; Stanford HAI on AGI; OpenAI on superintelligence.
