Just a few weeks in the past, I wrote a constructive article on utilizing AI to generate names in your startup: How one can Use AI to Generate Names for Your Startup.

Now here’s a contrarian perspective, or extra precisely, a view that discusses a few of the important shortcomings of AI that that you must acknowledge, perceive, and modify for. In the event you don’t account for these shortcomings, you’ll have a more durable time utilizing AI to generate names.

I began by asking ChatGPT a quite simple query — “What’s AI actually dangerous at?” Right here is the response:

Whereas synthetic intelligence (AI) has made important developments in varied fields, there are nonetheless sure duties and areas the place it tends to wrestle. Listed below are some areas the place AI could be thought-about comparatively weak or faces challenges:

1. Widespread sense reasoning: AI methods usually lack the flexibility to know and apply frequent sense data, which people purchase effortlessly. They could misread or misrepresent sure situations that appear apparent to people.

2. Contextual understanding: AI can wrestle with comprehending context, particularly in terms of language and ambiguous conditions. Understanding humor, sarcasm, or cultural nuances could be difficult for AI fashions.

3. Moral decision-making: AI methods lack an ethical compass and wrestle to make moral judgments. They depend on the information and directions they’re given with out inherent moral reasoning. This limitation raises issues when AI is tasked with making important choices or working in delicate areas.

4. Dealing with advanced and novel conditions: AI fashions are sometimes skilled on historic information, they usually could wrestle to deal with conditions or issues they haven’t encountered throughout coaching. When confronted with unfamiliar situations or novel duties, AI could produce unreliable or faulty outcomes. Emotional intelligence: Understanding and expressing feelings is an space the place AI falls brief. Whereas AI can acknowledge patterns in facial expressions or analyze sentiment in textual content, it lacks true emotional understanding and empathy.

5. Creativity and originality: AI excels in sample recognition and producing outputs primarily based on present information, nevertheless it struggles to exhibit creativity or produce really authentic concepts. AI can mimic or remix present content material however could lack the depth of human creativeness and innovation.

6. Bodily dexterity: AI sometimes operates in digital areas and lacks bodily presence or dexterity. Advanced motor abilities, fine-grained manipulation, or duties requiring bodily agility are nonetheless difficult for AI methods.

It’s essential to notice that AI is a quickly evolving area, and researchers are continually engaged on addressing these limitations. Nevertheless, as of now, these are areas the place AI is mostly thought-about to be weak in comparison with human capabilities.

I applaud ChatGPT for being so trustworthy!

This response revealed a number of methods the shortcomings of AI would affect the event of names.

One purpose many individuals use AI for title technology is that it is vitally straightforward to generate many concepts. So many concepts. The visible picture of making an attempt to get a drink from a full-on firehose involves thoughts.

And but, so a lot of these AI-generated concepts are both silly, generic or already trademarked. AI doesn’t have an efficient filter for high quality, strategic match of names, or trademark availability. A human wouldn’t recommend a reputation like “LeafyLung” for a pure cough syrup, however AI did in a latest mission.

So don’t suppose it’s “really easy a caveman can do it®.” Though AI can generate a large checklist of names, you continue to have a variety of work to do to establish these names that make sense and ensure they’re accessible from a trademark standpoint. The sheer variety of names results in an unimaginable quantity of labor.

Your model exists in a sure context at a second in time. AI can’t interpret your model’s emotional, social, and cultural contexts. People are good at this however AI, not a lot. You will need to assist AI perceive that context by superior prompting.

Asking AI to “generate names for my French Restaurant” shouldn’t be useful. You’re going to get generic names that aren’t targeted on the specifics of your mission.

As an alternative, strive offering the correct context for AI to make use of in title technology, which ought to yield extra related outcomes. Right here is an instance:

“Act as knowledgeable naming professional with nice ability in growing names. Give you ten names for a French Restaurant in downtown Philadelphia that excels at farm-to-table cooking in a easy “nation cooking” fashion. Whereas different French eating places are both casual bistros or high-end haute delicacies, our restaurant is the in-between different with easy, on a regular basis French delicacies. As a result of our location is Heart Metropolis, we count on our clientele to be workplace staff and people who find themselves occupied with a meal earlier than a present or a sporting occasion. Please give the which means behind every title and your rationale for recommending every title.”

With out straying into the quicksand of “moral branding” or “woke advertising and marketing,” let me simply say that that you must apply an acceptable filter to your outcomes. AI will usually generate concepts that your goal market would possibly think about offensive. Purchaser beware.

In case your naming problem requires “out-of-the-box” blue sky pondering, then you’ll have to work laborious to get AI to give you the results you want. Don’t get me incorrect, AI can establish inventive angles, however in the long run, AI is a regurgitation engine that is excellent at remembering a variety of issues and placing these issues collectively in a method that is sensible.

Typically AI applies creativity in unusual methods. For instance, a latest New York Instances article indicated that AI usually makes stuff up: “The brand new AI methods are “constructed to be persuasive, not truthful,” an inner Microsoft doc stated. “Because of this outputs can look very lifelike however embrace statements that aren’t true.”

Inventive pondering in producing names requires a stable strategic base, an innate curiosity concerning the world wherein your model exists, and a willingness to discover other ways to specific your model identification. AI wants a variety of assist to perform this. For instance, did you ever discover that human arms are sometimes distorted in an AI-generated image? That’s as a result of AI has inadequate “expertise” in seeing human arms and can’t correctly render them (most images don’t embrace arms). Your Goal Market has important human expertise along with your model (or that of a competitor). It’s essential to faucet into that humanity to get the very best title.

Doesn’t apply.

NOTE: I’m not calling anybody silly. However I’m declaring that silly use of AI will generate horrible outcomes. The previous adage “Rubbish In, Rubbish Out” applies in spades to AI.

AI can do an exquisite job at producing names if, and provided that, you’re good sufficient to ask the appropriate questions on the proper time to ship AI off in the appropriate course, then tweak the method as you start getting outcomes so AI generates higher names, then use strategic discretion to guage the outcomes of the brainstorming.

It helps to have a stable strategic basis in your naming train earlier than you begin producing names so you possibly can level AI in the appropriate course and hold it on monitor. Moreover, it makes judging the large variety of concepts AI will generate a lot simpler.

If you’re unclear about growing a stable strategic basis for title technology, my DIY Naming ebook covers this course of intimately.