Bizarre AI Research, Perplexity Ad Blunder, New RWS Hires

24 Oct 2025 · 24 min · 7 chapters

Ask about this episode

Ask anything about it. ChatGPT or Claude reads this page and answers with the times it was said.

Connect VO and ask about every podcast you hear, including the moments you saved. Add to ChatGPT · Add to Claude

In short

SlaterPod quick news for localization/translation industry: (1) copy-pasting localization manager job ads, comparing OpenAI vs Perplexity; (2) RWS Q4 strategic leadership hires; (3) critique of a UK DESNZ-commissioned report claiming AI translation is 1,000x more energy efficient than human translation; plus podcast recommendations.

Guests

Esther and Florian (hosts). No external guests mentioned.

Guest backgrounds

Esther and Florian are involved in advisory/consulting for language services and hiring; they discuss representing clients in M&A and advising on localization leadership roles.

Key claims

Perplexity’s localization manager ad is nearly line-for-line copied from OpenAI’s, including responsibilities and qualifications. RWS hires Steven Lam (CFO) and Michael Wayne (Head of Media & Entertainment, from Papercup). DESNZ/Europe Economics report compares AI vs human translation energy use using questionable assumptions.

Notable examples

50,000-word translation energy figures (<0.05 kWh AI vs 29–121 kWh human) and “copy paste” wording like “collaborate with GTM product ops… ensure the right surfaces.”

Written by AI. May contain mistakes. Listen to the episode to check what was said.

Chapters

Tap a time to open that second in VO

Willem Language Services Sale

0:19 to 1:28

Discussion on the recent sale of Willem Language Services and the advisory process.

“We are going to talk about copy pasting of localization managers, number one.”

Perplexity's Job Ad Controversy

1:28 to 4:28

Analysis of Perplexity's job ad, comparing it to OpenAI's and discussing implications of copying.

“So Willem being based out of Budapest, sold to a newly created brand In Any Language.”

RWS Leadership Changes

4:28 to 13:30

Overview of RWS's strategic leadership hires and their potential impact on the company.

“I want to talk about something that was really funny.”

RWS New Hires and Future Directions

14:00 to 17:48

Discussion about new leadership hires at RWS and their significance for the company.

“On the CFO role, there are a couple of comments just about the background and maybe pointing at the direction of the company in future.”

Bizarre AI Translation Research

17:49 to 22:09

Analysis of a controversial research report claiming AI translation is 1,000 times more energy efficient than human translation.

“Now, shall we transition to the most bizarre piece of research in the history of translation?”

Critique of AI Research Premise

22:10 to 23:04

Critique of the logic and implications behind the AI translation energy consumption study.

“the energy consumption associated with these professions, not pitting AI against humans in certain professions.”

Podcast Recommendations and Events

23:05 to 24:09

Recommendation of another podcast and announcement for an upcoming event.

“Short, I want to draw your attention to my new favorite podcast by Dwarkesh Patel and he had two amazing episodes recently.”
Hear the part that matters, and keep it.Open this episode in VO. Double tap your headphones to save a moment as you listen.
Get VO free

Transcript

Automatic transcript. May contain errors.

0:00Esther Bond:They looked at translating 50 ,000 words and apparently AI translation uses less than 0.05 kilowatt hours. And the human translator uses between 29 and 121 kilowatt hours.

0:19Esther Bond:And we're back with SlaterPod. Hi, Esther.

0:21Florian Faes:Hi, Florian.

0:22Esther Bond:Quick news episode to catch you up. We are going to talk about copy pasting of localization managers, number one. Number two, RWS announcing strategic hires in Q4, some big hires there, and visual data doing the same. And then I want to talk about the most bizarre piece of research on AI translation you have ever seen. And that was our actual headline, which is something we don't usually do. these types of headlines, but this was indeed the most bizarre piece. And then finally, I just want to recommend a couple of podcasts you do have to listen to, so a podcast with podcast recommendations. But first, Esther, big week for Willem in Hungary.

1:11Florian Faes:Yeah, we wanted to congratulate our client, Willem Language Services, on its sale. So yeah, very pleased to have represented Willem in the M &A sell side process. The Willem recently, the announcement this week was that it was sold to a company called In Any Language. So Willem being based out of Budapest, sold to a newly created brand In Any Language. So yeah, very exciting for everyone involved to Tunde Galberi, the CEO of Willem, and then also Jung Min-joon, the CEO of In Any Language, as well as us for having been involved.

1:51Esther Bond:So how were we involved? Let's tell people.

1:54Florian Faes:Yes. Well, so we were the advisors. We were the, strictly speaking, joint exclusive advisors. That means that we work with our partner firm of Area Advisory and we support the seller, so representing the seller in this case, end to end in its sale. So from pretty much the minute they decide that strategically a sale or an acquisition is the right thing for them. We got involved. So that's sort of helping to prepare all of the documentation, the materials up front, helping to define the positioning before you actually announce to the market that you are pursuing a sale. So right from the preparation phase, we then go out and contact possible buyers that we define in the preparation phase.

2:48Florian Faes:Do a lot of admin and back and forth at that point. So reminders, chasers, NDAs, Q &As, call scheduling, all of that. And then also really walk the seller right through to a closing. So that means obtaining an offer or multiple offers, then helping with negotiations, doing the virtual data room and supporting due diligence and then yeah doing all the updates and hand-holding right through to closing.

3:22Esther Bond:It's a very controlled process and it's when you're saying announce to the market this is a very tightly controlled and also highly kind of confidential process as well so you know you'd only reach out to people that were obviously approved by the seller so yeah so if you have an LSI, if you have an LTP and you're looking to sell that company, looking to exit, looking to partner with somebody else, come to us and we'll walk you through it. So this week with Willem, this was our, what was it? Our fifth or fourth? Yeah. Fifth successful exit that we helped manage for a client. Yeah, this year in 2025.

4:05Esther Bond:Yes. In 2025, yes. So that's...

4:08Florian Faes:In 10 months or whatever. Yeah.

4:10Esther Bond:Quite the track record. Definitely, yeah.

4:13Florian Faes:It's been a busy year. Busy and successful so far.

4:16Esther Bond:Busy and successful. Cool. All right. If you want to talk, head to the website, click on schedule a call and, you know, you can schedule a call with Esther and we'll look forward to hearing from you. All right. Let's go back to the news here. I want to talk about something that was really funny. I think it's just hilarious. And on multiple fronts. also wrote about it in the weekly newsletter on Friday. But for those few of you who listen to the part but don't read the newsletter, I want to make sure you know about this. So basically, we must have spoken about this OpenAI localization manager ad a couple of parts ago where OpenAI, you know, kind of the irony like OpenAI, everybody is looking at them to automate and use language AI to automate localization.

5:02Esther Bond:But now they were hiring, I think they put an ad out about a couple of months ago that they were hiring a localization manager. And so this week or last week, we came across a job ad by a company called Perplexity.

5:18Esther Bond:They have a browser also like OpenAI now does since this week, but they have a browser. They're like just one of these, I guess what you call labs, but their product was a browser and you can ask all kinds of questions and they do all other stuff. Have you ever used Perplexity?

5:35Florian Faes:I've probably tested it or played around with it, but it's not one of the ones I would use on a regular basis, no.

5:40Esther Bond:All right. So AI lab called Perplexity also looks to hire a localization manager. And so we're like, okay, that's interesting. So there's a pattern. You've got OpenAI, you've got Perplexity. And then we looked at the job ad and we were like, okay, I don't know. That sounds like oddly familiar. So when we were starting to look at the individual responsibilities, It was obviously interesting to see, well, is there any difference to OpenAI? It turns out there's very little difference. And if you stick the two ads into, well, ChatGPT and you ask it to compare, let me read what ChatGPT tells you. It says, apart from one or two newly added or expanded sentences, the structure, order, and language match OpenAI's description nearly line for line, indicating clear text reuse rather than coincidence.

6:26Esther Bond:And so let me spin this up for those of you are listening or watching this on YouTube. It's just, it's kind of funny. So yeah, let's go through this. So for example, on the responsibility side.

6:42Florian Faes:Which one's which? So you've got...

6:44Esther Bond:Sorry, on the left here is perplexity, on the right is OpenAI. Okay.

6:49Florian Faes:Okay.

6:50Esther Bond:So for example, you have things like responsibilities, manage, now which one is, let me look at this. It just starts out by drive localization strategy and execution

7:07Esther Bond:with alignment to product and other orgs and OpenAI is like drive localization strategy and execution across marketing services. So at least one difference is with perplexity you're reporting to product with OpenAI you're reporting to marketing apparently. But then you have like things like collaborate with GTM product ops, partnerships, legal and regional teams to prioritize languages, align on launch time and ensure the right surfaces are localized for the right audiences. That was perplexity and now OpenAI collaborate with GTM product ops, partnerships, legal regional teams to prioritize languages, align on launch time and ensure the right surfaces or localized for the right audiences.

7:50Esther Bond:It's literally the exact same thing.

7:54Florian Faes:It's very specific wording. I think like surfaces is like not really something, I don't know.

8:00Esther Bond:It's the, I mean, it's so copy paste. Yeah. And also collaborate with GTM. So when OpenAI says collaborate, go to market, product ops, partnerships, legal and regional teams, and Papraxia says collaborate, go to market, product ops, partnerships, legal and regional teams. Do you guys have the exact same structure in your business? Okay. I mean, I could go on. Same with the qualifications. Like, for example, OpenAI says strong knowledge of localization workflows, tools, TMS, CMS connectors, API, and infrastructure. And yeah, copy paste the exact same thing in the perplexity ad. Yeah. What are your thoughts?

8:39Esther Bond:Copying an ad, good or bad? Discuss.

8:42Florian Faes:I don't mind.

8:44Esther Bond:You don't mind?

8:44Florian Faes:Personally speaking. I mean, let's say, I mean, I've seen similar things in the past. For example, you know, with recruiters, if you're, recruiters are often hiring pretty much the same position across multiple companies. So, you know, when they're maybe doing their own version of an anonymous job ad, it's going to be quite high level the same thing. The idea being you'd maybe put multiple candidates forward for, sorry, the same candidate forward for different roles. so you are essentially looking for the same profile and the same person really but when it is the business or you know HR let's say putting together a job ad to advertise their company or trying to attract candidates to work at their company you would probably expect a little bit more tailoring nuance to it.

9:36Florian Faes:However, if this is an entirely new role and function potentially within both companies, then yeah, why not? Saves on the admin, saves on the writing.

9:49Esther Bond:I'm actually a little torn on this. On the one hand, you're like, sure, if OpenAI spent a ton of time and thought on this ad or spent some time and some thought on this ad, like if I'm in a similar business like you said, like why wouldn't I copy? Now that said though, if I was a candidate and be like, well, you guys are copy pasting the exact thing, like obviously it can't be the exact same role. And like when you're talking about go to market legal and all these other things, like I would feel that they're not taking this hire super seriously. I think they'll still get enough candidates.

10:28Florian Faes:I've got to say. No, no, no, no. No, but what I mean, like, okay, now then you do join and, okay, let me try to rephrase this.

10:38Esther Bond:So you have an internal problem, you have an internal challenge, right? Yeah. And then you put together a job ad that reflects that challenge. And the outcome is an original job ad, which is customized to the challenge at hand. And then as a candidate, like, well, I'm obviously joining a company that's thinking its challenges through and I am now filling this role. And this feels like, okay, we need a localization manager. Let's get some ad. And then, okay, I join. And whatever it was in the ad is obviously not really the reality. I'm just joining and I'm expected to do something, probably not all the things that we're in the job ad.

11:23Esther Bond:So I feel a little bit like, are you guys like serious about this? Like, what do I need to do?

11:28Florian Faes:I know what you're saying. I think I've seen these sorts of roles in the past where a company that previously hasn't had a localization function or any kind of localization personnel, they know that they need something, but maybe they don't know what the ins and the outs of that role might be. So the point is, get the person that seems like they can deliver this or has done similar things in the past, get that person into the role and that person will kind of need to build everything around them. So, yeah, maybe there isn't anybody, the person responsible for the job ad or the hiring manager might not, I mean, probably is not expert in localization or language.

12:09Florian Faes:So I know what you mean, like the signals that are being sent are very much, we need a person who knows about this. But that being said, I think they will find that person.

12:25Esther Bond:Yeah, they will find that person. They will find that person, I'm sure. Especially with the salary attached. Yeah, like 200K a year. Yeah, I mean, it was like even an hour after this was posted, that had like 100 applications already or something like that. Well, anyway, good luck, Perplexity. And if you have questions around how to hire a localization manager, you can always call us. We have a consulting practice that's more than happy to help you. And yeah, if you do end up being the selected localization manager at Perplexity, we'd be happily... Join the podcast. Join the podcast, speak at SlaterCon, speak at SlaterCon remote, which is coming up in December, early December.

13:09Esther Bond:Now, I wonder if RWS copy pasted a rival LSI's ad when they were hiring their new CFO. I'm going to say no. Probably not. They probably didn't put out an ad. They probably paid some like, you know, expensive recruiter. Executive search. 30 % of the CFO's salary. No, anyway, let's move on. RWS announces strategic Q4 leadership hires, So two senior hires at RWS, what do we know about this?

Read the full transcript

13:37Florian Faes:Yeah, so one is a new CFO, Chief Financial Officer, and one is a new Head of Media and Entertainment. So we have Steven Lam joining as the company's new CFO and Michael Wayne joining as the Head of Media and Entertainment. So two pretty high profile visible roles there that are joining RWS. On the CFO role, there are a couple of comments just about the background and maybe pointing at the direction of the company in future. The CEO of RWS was saying that Stephen has a successful track record in financial leadership of high growth and listed global businesses, as you would expect for somebody entering a CFO role in a listed company.

14:26Florian Faes:also that he has a track record of executing M &A strategies among other things which will be invaluable in supporting our success and then on the head of media and entertainment role just a bit of background on that the the person Michael is based in Los Angeles obviously the heart of those kind of services and he will lead the company's media localization division which includes all the services you might expect like dubbing, subtitling, AI dubbing and comprehensive content adaptation. Michael actually comes from most recently Papercup so the AI dubbing company that RWS acquired, well they acquired the IP right back in well earlier in 2025 I think it was.

15:21Florian Faes:So yeah, he was the chief commercial officer at Papercup, whose IP was acquired by RWS and now joins RWS in the role of head of media and entertainment. And they say it's part of an accelerated investment strategy into video localization with a view to combining both the AI expertise and human creativity as well. Makes sense, right?

15:50Esther Bond:Ben Fez's background, the CEO, his background, he was with YouTube for close to a decade, I believe. So he has intimate knowledge about what's required to succeed on YouTube. And so this is something they're going after.

16:09Florian Faes:He probably feels quite comfortable pursuing that particular segment. Yeah, within that segment. You're right.

16:16Esther Bond:And with Papercup, didn't they go to Scale AI? I think the team went to Scale and the IP to RWS.

16:21Florian Faes:That's right, yeah. So when I was saying, oh, he joins as part of the acquisition, I suddenly thought, well, no, because the IP, it was a split. So you had the IP and the technology, which went to RWS, and you had the team, at least the announcement at the time, was that the team was joining Scale AI.

16:39Esther Bond:All right, and we had another senior hire, this time from Visual Data, closer to where you are.

16:45Florian Faes:That's right, yeah. So this is an announcement from Visual Data that they have a new SPP of worldwide localization. The person is Maz Aljumali, who is going to come on board with Visual Data, responsible for localization services such as subtitling and dubbing. He's based in London and actually comes, I think, most recently, I believe, from a role at Zoo Digital, where he was for quite a while. prior to that also with sort of rivals competitors similar media localization companies bti studios which no longer well is now part of sdi and then iuno and then and then deluxe so yeah you see a lot of movement typically between these kinds of media localization companies but he's going to be based in london or is based in london and will focus on client engagement strategic partnership which is obviously a big topic at the moment, and growth in key content markets.

17:49Esther Bond:Now, shall we transition to the most bizarre piece of research in the history of translation?

17:57Florian Faes:Seamlessly, yes. Let's do it.

17:59Esther Bond:Seamlessly. All right. And responsible for this beauty is the UK's Department for Energy Security and Net Zero, D-E-S-N-Z. And they commissioned a consultant they called Europe Economics to study digital services energy use. And there's a bunch of stuff in that. But one of the key use cases they featured was AI translation. And they claim, I mean, the outcome is that AI translation is 1 ,000 times more energy efficient than human translation. Okay, where to begin? So basically, they're looking at translating 50 ,000 words and apparently AI translation uses less than 0.05 kilowatt hours and the human translator uses between 29 and 121 kilowatt hours.

18:58Esther Bond:And the report calls this, that's why we picked up on it, that the whole report calls this the most dramatic contrast among all digital versus physical activities analyzed.

19:11Florian Faes:Okay, what's the office space? Because I'm struggling to visualize. It says the human translator works in a 160 square foot office. How big is that? What do you mean?

19:22Esther Bond:160?

19:23Florian Faes:160 square foot in meters. I don't know. It's like a smallish office, I suppose.

19:29Esther Bond:It's a smallish office.

19:30Florian Faes:Yeah.

19:31Esther Bond:It's not big. Sorry, that's what, 15 square meters or something. Okay, you're jumping right into one of the details.

19:37Florian Faes:Yeah.

19:39Esther Bond:Well, so if you look at this, so basically you are comparing, I mean, I think as the image for the article, we had like apples and oranges, but it doesn't even do it justice. Like it's just comparing, it makes absolutely no sense to compare these two things, like generally. Like, so if you think it through the end and you're basically implying that less energy, that replacing a human activity with something that's done by AI and you're putting it through an energy lens, like if you think it all through the end, like it becomes very philosophical almost like, all right, so why don't we just like replace everything with AI?

20:24Florian Faes:Why do humans need to play tennis when AI can play tennis?

20:28Esther Bond:Correct. Why don't we watch human tennis? It makes no sense. We could just, or like, you know, football, like, why do we have all these people go to a football stadium? We could just play FIFA all day. Like, you know, makes no sense. And I mean, we could give a gazillion other examples. So the premise makes absolutely no sense. And then if you, yeah, if you break it down to the translation level, which again, they're saying is the most dramatic contrast. Yeah, okay, so you're literally counting that human's office space, you know, the sandwich, the heating in the winter and the aircon in the summer.

21:06Esther Bond:I don't know. It makes no sense and you don't consider, I mean, then if you like go a little more nuanced and you kind of accept the premise, which is very hard if you accept the premise, let's assume we're accepting the premise, then you don't consider the quality, okay, and they also used some outdated speeds. I think they're taking like a 400 words per hour human translation speed, which you know, today everybody's doing post-editing anyway, so it would be much higher. Yeah. And then, yeah, some, what else do we have? Uneven energy accounting. They're saying, for example, corporate energy for AI providers excluded as immaterial.

21:48Esther Bond:Again, a human gets 150 square feet off his space. Anyway, we just wanted to draw your attention to this piece of research. We cover a lot of research and usually we try to be very neutral, but when you come across something like this, you're like, you can't be neutral anymore.

22:08Florian Faes:It seems like maybe a good angle would have been like, here's how AI is helping to reduce the energy consumption associated with these professions, not pitting AI against humans in certain professions. You see what I mean? Because it's like, say, AI is making translation more efficient because we're doing more words per hour effectively, you know?

22:33Esther Bond:I still struggle with the premise. I mean, that human translator, even if that human translator wouldn't be translating, they would be sitting in an office, I guess, doing something else. Yeah, you still got to breathe and eat whether or not you're doing translation. Yeah, they don't evaporate into space. I hear you. All right. Well, that was I think the third attempt to summarize this. I mean, once in the article, once in the newsletter and now on the podcast. So let's see. I think this one. Anyway, funny. All right. Short, I want to draw your attention to my new favorite podcast by Dwarkesh Patel and he had two amazing episodes recently.

23:19Esther Bond:So one was with Andre Kapathi, one of the most, to me, most informed commentators on AI and NLMs and then also with Sutton, I'm blanking on the surname. Anyway, Google it, the Sutton episode, S-U-T-T-O-N by Dwarkech. So just go on YouTube or on your favorite podcast app and do catch up on these two episodes. Amazing work there by Dwarkech. Really, really good to catch you up on kind of where we stand in the hype cycle maybe. It gives you a little more context to your personal context window. All right. And while you're at it, also register for Sliticon Remote happening in early December, wishing to see you there.

24:09Esther Bond:Have a good one.

From the publisher

Florian and Esther discuss the language industry news of the week, with congratulations to Villam Language Services on its sale to InAnyLanguage. Slator served as joint exclusive advisor with Maveria Advisory, representing Villam throughout the end-to-end M&A process.

The duo turns to Perplexity’s Localization Manager job posting, which they found almost identical to OpenAI’s earlier post, down to matching structure, order, and phrasing. They question whether copying such a specific ad shows a lack of seriousness or simply reflects practicality and efficiency.

Esther and Florian talk about RWS's new leadership hires: Stephen Lamb as Chief Financial Officer and Michael Wayne as Head of Media and Entertainment. Esther outlines how the appointments strengthen RWS’s investment strategy in media localization, dubbing, and content adaptation.

Esther next mentions that Visual Data has named Maz Al-Jumaili as SVP of Worldwide Localization, to lead subtitling and dubbing operations and strengthen client partnerships.

The duo wrap up with the UK government’s bizarre energy-efficiency study, claiming AI translation is a thousand times greener than human translation. They review the flawed logic, where the report assigns human translators the entire office energy costs while excluding AI infrastructure.

More from SlatorPod

All 39 episodes
Bizarre AI Research, Perplexity Ad Blunder, New RWS HiresSlatorPod · 24 min
Listen in VO