“Can’t clients just do this themselves with AI?”

Free webinar on Friday

On Friday, March 27, guest instructor Samuel Mowry will teach a one-hour free webinar, An introduction to LSP.Expert. If you’re still using spreadsheets for accounting and project tracking, LSP.Expert is a better way! It’s designed by translators, at a fraction of the cost of programs like QuickBooks. Come learn what it’s all about! This session is not a sales pitch; it’s an informational introduction, after which you can try (or not) the free trial of the software. Registration is free and includes the recording.

Also, it’s interpreting exam season! On April 16, I’m teaching How to prepare for (and pass!) an interpreting exam. Registration is $75 and includes the recording.

“I try to motivate myself to market, but then I think…can’t clients just do this themselves with AI??”

It’s a common fear these days. I’ll admit that I don’t share this fear, which makes me either the best, or the worst, person to write about it (we’ll see!!). How do I market my translation services, without feeling like I’m selling something that clients can get for free? Here’s what I’ve heard from a few colleagues:

• “Marketing these days feels dishonest, like I’m hoping clients don’t know that they could just use DeepL”
• “Good enough is good enough for most clients; they’re totally happy with an AI translation and won’t pay for a professional translator”

Here’s what’s not incorrect:

• Just as with any professional service, some clients are happy with “good enough.” Let them carry on with their tolerance for mediocrity while you carry on looking for clients who value excellence, service, and trust.
• For straightforward information in common language combinations, AI does a decent job. Again, that’s not what you’re selling; you’re selling a really well-written translation, done by someone who knows the terminology, the context, the cultural aspect, and is a really good writer (which AI is not).

DIY options aren’t a new thing

If you’re thinking that “software that does a similar task to a professional knowledge worker” is a new thing, like this all started with ChatGPT, it didn’t. Lots of professions exist alongside software that does some version of what the human professionals do. Sometimes, the software does virtually eliminate the professional: very few people these days go to a travel agent for something simple like booking a plane ticket. But in professions that require a high degree of knowledge, skill, or creativity, that doesn’t seem to be the case. For example…

TurboTax (DIY tax preparation software that many Americans use) came out in, wait for it, 1984, and the accounting profession seems to be holding up fine. QuickBooks? 1992. LegalZoom (a U.S. service that sells template legal documents for things like making a will, or selling your house): 2001. Various pieces of website creation software have been around since the early 2000s as well. Yet, bookkeepers, attorneys, and web designers haven’t been eliminated. There honestly isn’t enough money in the world to convince me to do my own taxes; the first business task I outsourced was accounting, and I’m never going back!

Likewise, many service providers do something that you could do yourself if you have the time and the desire. During COVID, I learned how to cut my own hair by watching YouTube videos. It actually looked OK (not just my opinion, validated by several friends!). But a few years ago, I went back to the salon; my hairdresser does a better job with less hassle. The same is true of tons of professional services: we’re hired to do things that people are not incapable of doing themselves, they just don’t want to or don’t have time.

Then there’s translation quality

This newsletter doesn’t have enough room to dissect the issue of AI-generated translation quality. But let’s say this, because it’s what I tell my clients.

• DIY can be a decent option for a low-stakes, information-only document that will be used only in-house (“Kind of/sort of what does this say in our language”).
• AI often sounds great, when it’s totally incorrect, and it does crazy stuff. A client came to me because they suspected that a translator had used machine translation. The clue? The name of the Swiss town “Aubonne” had been “translated” as “Augood.” No kidding.
• Raw AI output is a wildly bad idea for anything that carries a legal, reputation, or brand risk, or where consistent terminology is important. A lot of my clients use AI-generated translations of conference presentations. You’ll see things like the English word “capital” (in the financial sense) translated in three different ways, in the same paragraph. AI gleefully spits out, “capitaux propres,” “fonds propres” and “capital,” or it will use the same French word for two different concepts in English (“financement” for both “financing” and “funding,” for example). I’m sure other languages have these examples as well, and it’s not what you want in your annual report.

Many freelancers’ issue? Selling the wrong thing

In my opinion, many freelancers have a deeper issue, which is that they’re selling the wrong thing. Recently, someone commented on a LinkedIn post I wrote (I’m paraphrasing here), “So many translators go on and on about how complicated and difficult the work is, when they should really be emphasizing how much easier the client’s life will be when they don’t have to deal with translation anymore.”

Bingo. Clients want a solid translation. You can’t be a nice person and an incompetent translator and expect to retain high-paying clients. But you can and should realize that most clients hire you to solve a problem and take something off their to-do list. You want to emphasize:

• Peace of mind: Once the client hires you, they no longer have to worry about their translations
• Risk reduction: No more wondering what’s in that AI-generated translation
• Trust and confidentiality: Their document won’t be blasted out in a “come and get it” e-mail to 50 translators
• Confidence and reputation: The documents you translate can stand up to legal, reputation, and brand risks; AI can’t stand behind what it produced, because it’s not human

Also: Getting clients out of the AI slop

In recent months, I’ve also landed some new writing clients who hired me specifically to get them out of the AI slop. Ghostwriting for the CEO; writing the company newsletter; doing case studies and client interviews.

I’m a little skeptical of claims/hopes that the pendulum is swinging away from AI. I think that there are just too many clients who prioritize price and speed over trust and quality. But I do think that there are many, many clients who have certain tasks for which they will not use AI, and they’re willing to pay quite well for an explicitly human touch.

I hope these tips are helpful!