For those who fear or denigrate generative AI.

There is talk in the press of AI being used as a scapegoat excuse when corporations are actually paring down their workforces for other reasons.

3 Likes

True, and the gen AI of 10-15 years from now will likely be massively different than that of today.

1 Like

I don’t mind being frank and honest about my job. I also will be the first to admit that I don’t have hard numbers or anything more than just my own personal experience to go on, but here goes…

I’m a translator, editor, copywriter and localization specialist. This means that I help companies adapt their content - magazines, video games, websites, books, etc - to foreign markets. I specialize in French and Italian content, and my core expertise is obviously in the English language. After being ā€œlet goā€ from a major video game company in Vancouver (they outsourced my department), I went freelance and set up my own business. I have had the good fortune to have a steady income with loyal clients for about 15 years now. My main area of expertise is actually video games, and my job has involved the creative part (writing) as well as the technical part where the content needs to be integrated into the code, although this mainly in a consulting capacity these days.

Over the last couple of years, several clients have said that they are using AI now for the translation and editing part of their content pipeline. Other clients have pivoted to using AI to do an initial translation and then passing it on to native speakers to polish it. This is where I insert the comment about a polished turd still being a turd :slight_smile:

I’m obviously in the translation community, and it’s been brutal over the last few years. Loads of translators and proofreaders are losing their jobs to AI. It’s no longer a viable profession for the vast majority. I’m faring better than a lot of people because of certain clients who really value what I do, and because I have a lot of experience compared to many. But this particular industry is one where there’s a wide-held belief that AI is as good or nearly as good as a human. It’s amazing how many people who don’t think AI is a threat readily admit that it’s a threat when I tell them what I do. ā€œAh yes, well I can see how translation and writing is at risk.ā€ Why, because it’s easier than other art forms??

AI is actually very good at many forms of translation and writing, at least on the surface. But I think the excitement over it is still overblown. It’s rather bland, derivative, still needs a human to make decisions, and often gets things wrong. So, I actually don’t see my industry as being too different from any others. It’s just that most people outside the industry just think it’s a matter of matching A in one language to A in another language, rather than the actual process of localization, which involves nuance, cultural adaptation, sensitivity to laws/rules/norms, idioms, etc.
One of the big projects I work on is full of jokes and silly wordplay aimed at young teens. But those jokes need to be adapted to English, and so there’s a dual skill of understanding the original joke, and then creating a new one suitable for the target audience. Sometimes my job is literally to come up with lots of dad jokes and wordplay that would make Dr Seuss proud!

But guess who decides whether to use AI or a human? Yes, the bean counters, and AI is an excellent way for them to save money.

I personally believe there will be pushback at some point, and customers may eventually demand human-created content. They already are doing this with certain video game development jobs, such as concept artists, voice actors, script writers, etc. But I fear the pushback will come too late for me in my role because it might not come until near the end of my career.

Anyway, this was longer than I expected, but I just wanted to share my own personal experience. Yes, AI is taking jobs in my industry at least. It may not be permanent over the long term, but my industry will almost certainly shrink and be de-skilled. Many companies now look for someone reasonably competent in English to just tidy up their AI-generated content, instead of recruiting a language expert with decades of experience. It’s ā€œgood enoughā€.

10 Likes

That’s an interesting perspective, thanks for sharing. Interesting because the video games industry has a history of poor translations and has evolved a lot over the years because it is such an international industry.

In the 80s and 90s the industry was mostly led from the Japanese developers and translations to English were notoriously bad. ā€œYou spoony bard!ā€, ā€œThe truck have started to moveā€, ā€œmaster of unlockingā€, those are just examples off the top of my head ranging from the 80s up through the mid/late 90s. A lot of those weird translations are because people just tried to literally translate from Japanese to English and budgets and international communication were minimal. But over time the industry put a little more effort into things and realized you can’t always just do a literal translation from one language to another, sometimes you have to rewrite things because of cultural reasons or humor that wouldn’t make sense in another language.

AI-first translations make me think the industry is going back to how it used to operate, with minimal resources put into those little details. But video games aren’t a niche thing anymore, and while those old mistranslations were funny remnants from a scrappy industry era, I don’t think your average consumer is going to find it funny when it’s from a huge budget AAA title in today’s industry.

1 Like

Yes, sadly that’s the case, especially for the AA and indie developers on limited budgets. I used to work for an AAA studio making one of the biggest selling titles of all time. Localization was often treated as one of the last things to prioritize, whereas the graphics, technology, motion capture and gameplay had tons of resources pumped into them. But they started to change their ideas when it became clear how many users there were in other countries and how much those users wanted the game in their own language with well-known personalities from their country.

Still, it was always a big marketing decision whether to just ship the game into that country in English or to invest in local voice talent and translators. Always cost/benefit.

Basically, translation/localization has always been one of the least glamorous aspects of video games, and most of the development team have no clue about it. So it’s often an afterthought.

You just need to watch the credits to see how far down the list people in my industry are, even though we are sometimes responsible for lines and moments that people love and remember forever. But I guess that can apply to the terrible translations/writing too, as you pointed out!

1 Like

I don’t even think the average person realizes how much of an art translation is. Even people who fluently or natively speak 2 languages often still have the perception in their head that translatation has a correct solution and the role of a translator is to arrive to that solution. So they see no issue with throwing the text into a machine and having it spit out something that ā€œtechnicallyā€ means the same thing as the original… most of the time.

2 Likes

I have met many bilingual and trilingual people who are terrible translators, and conversely some very good translators who struggle to speak another language. People are often surprised when I tell them that my main skill is writing English. They always assume that I must be incredible at French and Italian. And while I’m obviously proficient in those languages, I’m nowhere near bilingual. I rarely speak those languages - just when I’m on holiday usually.

But a good translator should be excellent in their native tongue. Your native-language writing is actually what people read, and the ultimate goal is that they have no idea that it’s a translation. It needs to be clear, entertaining, interesting or whatever you need to convey, and very natural so there’s no hint that it was originally another language. There’s actually a term in the industry called transcreation, which means you create entirely new text or messaging with the original content merely used as inspiration. Ad agencies basically do this when launching a new product in a new country.

Anyway, I don’t want to big up what I do too much. It’s just a piece of a larger puzzle in content creation. The parts I enjoy most are the creative aspects, and creativity is what I want AI to leave alone. Humans are good at it, so we don’t need algorithms to do it.

3 Likes

I used to work for a company doing speech-to-text automatic speech recognition. This used to be an area of active study and development.

Then OpenAI came along, and needed to gobble up YouTube, so they created a speech recognizer. To them, it was just a means to an end, they weren’t even interested in the problem. That was the Whisper model. That model entirely killed the industry. Whisper is free, accurate, speaks 100 languages, and can output text in a different language than the speech. It was ridiculous.

It didn’t impact me personally. The company I worked for had bigger trouble than our four-people team, and I was on my way out anyway. But the market of speech recognition software completely collapsed afterwards.

2 Likes

Yep, I can imagine. In the early days of my freelance business, I sometimes took transcription jobs. It was low-paying but if you could type fast, you could earn a bit of money. But it quickly became very clear to me that there was no future in that particular segment. Automated speech recognition was always going to win that one, and I have much less of a problem with that. Transcription is much more about matching X to X and Y to Y with very little creativity needed, if any.

I’m not so biased that I can’t see where using AI and language models makes sense. My problem is when profit is prioritized above all else. We shouldn’t be shoehorning it into everything just because we can. And other than concerns of actual productivity and quality gains, I do believe we have an ethical responsibility to approach it cautiously and think of the larger impacts on society, the environment, the economy, our future generations. The fact that it is being led by Big Tech with very little in the way of regulation is very concerning, and I don’t share the optimism of those who think that the bubble will burst and everything will sort itself out just fine. We have not just created a loom here.

2 Likes

Did you see this mesmerizing exchange today, between the matplotlib maintainers and an OpenClaw AI agent?

We live in a strange, alien world.

4 Likes

One of the problems with this, and the conclusion that we should have additional regulation is this: as technology becomes more powerful, it creates higher mountains for people to climb. In other words, it creates more and more lucrative opportunities that some people will do anything to take advantage of, and the average person will never have a hope of competing with the minority that have mastered taking advantage of new technology. Regulation cannot hope to keep up because it takes months or even years for regulations to be passed, whereas a new technology can be created and used in new ways in seconds.

What I think is that at some point, technology will reach such a high level of power such that regulation will be powerless to really control its development and use.

I think we should not be so concerned about productivity in the first place…

I agree with this, and I don’t think regulation on its own is enough to strike the right balance. But we still need regulation to add friction and help users/consumers in a rapidly changing world.

The technology is already out in the wild, so I don’t think we can close Pandora’s box now. I have an idealistic vision of what I want, as I’m sure you do too, but I also realize I need to be realistic about what can be achieved.

2 Likes

Interesting Soumaya Keynes interview with a China academic in the Financial Times. Looks like China’s next massive overcapacity, price collapsing issue – after solar, EVs, etc. – could be AI.

Yanmei Xie:

ā€œLet’s talk about this real-time example of China trying to build this ā€œAI Plusā€ economy. So last year, about August, the central government issued another action plan called the AI Plus Action Plan. The idea is that you sprinkle the magic dust of AI on everything — on production, on daily life, on transportation, on governance, and then that can spark productivity, stimulate growth, stimulate consumption, right? So because there was central government edict, pretty much every local government then came up with their corresponding AI action plan. So every government wants to have an AI industry in their province or in their city, right? And attached to these action plans, they have a target of growing AI industry by X year. And then attached to the target, they have subsidies — probably cheap credit, cheap land, tax rebates — to companies who claim to fit into this AI Plus paradigm, right? So overnight then you have tens of thousands Chinese companies just rush into this industry. Some are real AI industries, real AI companies, some are not. But nonetheless, they all claim to be these AI companies, AI plus companies to claim these subsidies and tax rebates, right? And overnight you have duplicated industries across the country, across provinces and they have supply overshooting demand.ā€

https://www.ft.com/content/10610138-d266-417f-bda6-cbf5bc9d6851

Meanwhile:

2 Likes

Note that private companies in capitalist countries are doing this without any central gvt intervention.

If you don’t, you are perceived as a dinosaur, even if you have a sensible approach.

Yes, except perhaps that this can last only as long as the bubble inflates and/or investors are willing to set fire to cash/credit in hopes of a longer-run return.

I’m probably late to this but… or should I say butt:

5 Likes
2 Likes

Sloppywood :rofl:

2 Likes

It seems that several top developers / designers are leaving their AI companies, recently.

https://thehill.com/newsletters/technology/5734448-musks-ai-firm-loses-cofounders/

1 Like

In the last few years, we’ve seen a number of high profile people joining AI companies. Well, I’d join an AI company, too, for a few million Euros. Now that their stock options mature, they leave again.

2 Likes