The Sword and Laser discussion

We discuss the return of Jim Butcher’s Dresden Files and Frank Herbert’s posthumous SFWA Infinity Award, Audible’s AI narration plans and Amazon’s cancellation of The Wheel of Time—while nominating hamburg steak and vegan enchiladas and sharing listener stories from Matt Dinniman’s event. We round out the show with in-depth discussions of The Lives of Tao and Mickey 7.

https://www.swordandlaser.com/home/20...

I can’t believe Tom admitted to being a pod person! 😮

I have to say, for me Mickey 7 didn’t feel like a one-off book that then had a sequel tacked onto it. For me the ending felt like a suitable pause rather than a full conclusion, and the sequel felt like a conclusion to a duology.

Ruth wrote: "I have to say, for me Mickey 7 didn’t feel like a one-off book that then had a sequel tacked onto it. For me the ending felt like a suitable pause rather than a full conclusion, and the sequel felt..."

Me am feel same.

Antimatter Blues doesn’t feel like a contractual obligation, and I liked it just as much as Mickey 7. I see parallels in other SF works, such as Bujold’s Vorkosigan series, Niven’s Known Space or Chamber’s Wayfarers books, where there’s plenty of room to tell further stories set in this universe featuring an entirely different cast.

On the subject of AI narration: I was one of the Indie authors Amazon offered AI narration to. Probably one of many thousands. Anyway, I said no. Probably AI will eventually become a part of audiobooks and comics. I feel no need to assist.

I sampled some of the AI narration and currently it’s bad. But next year it will be on par with humans.

A recent estimate said AI will replace 10 million jobs in the next 3-5 years. That’s a lot and soon. AI could definitely do 80% of my job at the TV station. Robots would be needed for the other 20%. If the system powered up and down by itself and we switched from wireless mics to wired ones, AI could do 95% of my job.

The recent Chinese robots I saw on YouTube last week are dangerously close to being able to do everything a person can. Fortunately we’re still years away from them being smart enough to do anything other than preprogrammed tasks, but it’s scary-slash-interesting how far they’ve come in just the last five years.

I am not worried about AI.

My first job in the Entertainment Industry was doing Accounts Payable for a major indie. I quickly segued into Distribution Finance. In that timeframe an "analysis" was a data dump out of the mainframe by the IT Director, which we then added up by hand. Any other kind of analysis was a report followed by tick and tie. It took the whole department working in parallel to do any kind of major analysis.

Now everything is either available immediately from the ERP or on a short refresh into analysis databases/software. A single analyst can do what a whole department needed 35 years ago.

Finance departments have not gone down. Rather, they do more analyses and at a deeper level. I expect the same with AI. The work won't go away. It will just change.

John (Taloni) wrote: "A single analyst can do what a whole department needed 35 years ago."

This is the crux of the issue, right here. There are some jobs which can be outright replaced by automation. In the short term, there will still be the need for a top-level human to coordinate things, but the other 99% of jobs are gone. Sure, the manager is still there, but all the employees below that level are out on the street.

I recall an interview with the guy who created the first really useful spreadsheet program, VisiCalc. He said that during their demonstrations that accountants would be visibly shaking because the program could do in a second what took them days or weeks to do. I saw an estimate that VisiCalc and its various successors caused more than three million people to lose their jobs. A similar thing happened when automation took over switchboard operator jobs. In 1980 there were more than two million women employed in that occupation. By 2005 there were 0. Automated operators still suck, but every business uses them.

My main job in TV consists of set-up, take-down, and pushing buttons to switch inputs between cameras and computers. That third thing can be mostly done by AI right now, and it is the bulk of my job. The thing a computer *can’t* do is deciding who to prioritize when multiple people are talking, but I can imagine someone deciding to simply program it to go into split-screen when several people are speaking.

I’m constantly cleaning my Roomba, so there will be jobs for robot repairmen. For a while. And there will be jobs for data analyst managers. For a while. Eventually someone will narrow those gaps and it will be robots all the way down.

Re: AI narration, Tom is far more optimistic than me. I agree that most people who would rather listen to an actor than an AI, even if they were equally understandable. But I don't think consumer preference is enough to outweigh the financial savings of making a computer do it when it comes to Amazon's decision-making.

Let me put forth this hypothetical: Let's say the next James Islington book (Will of the Many was an Audible Exclusive) is in production and the narrator wants a salary bump since WotM did so well. Audible says "Nah, we'll just use AI."
It is the only audio edition available, and I listened to the last one, and because of its length and the way it's written (first-person linear narrative), it is perfect for audio and I'd probably never get around to reading the physical book. Even if the quality is worse, and even if I am morally against supporting AI, I will probably end up listening to the AI narrator, because it's better than not reading the book/supporting the author.
The number of people who would refuse to listen for moral or quality reasons would probably be like 5-10% tops, and if the collective interest in the series builds enough, they would likely make more money off of book 2 in its first year than they did with book 1 regardless, so there would be no real financial incentive to appeal to that group, because even though Amazon sold an inferior product, it would be good enough for most people to buy it anyways.

That's what makes AI so profitable: not that it's as good or better than humans. In many areas, such as customer service and entertainment, it's obviously worse. But it doesn't need to be as good or better, it just needs to be good enough for us all to put up with it, because the company profits even if we like the product less.
Who's going to compete with Amazon/Audible after all? Is some start-up going to compete with Audible with the business model "We will spend more money creating a product that's already available and sell it to a very small subgroup of people who are both 1. Hardcore audio readers, and 2. Vehemently opposed to any interactions with AI." Worst business proposal ever.

Amazon has no incentive to create high quality audiobooks, because they have no competition. If they can sell mediocre quality audiobooks for a fraction of the cost, who will stop them? With that in mind, I think it'll be maybe 2 or 3 years before half of the top selling audiobooks are AI narrated.

I'll start off by saying, I'm against AI narration as a replacement for humans. It's going to happen, but I'm not happy about it. I will always buy the human version versus the AI. If given a choice.

But will the AI be bad. It is now, but it's only going to get better and I would assume the best companies will still have a human sound editor tweaking the audio to remove any 'Uncanny Valley' oddities and to smooth out the quality. I expect it will be indistinguishable from humans within a few years.

AI does have its use. For books that will never sell enough to warrant the cost of a human narrator.

Also for better, on the fly, narration for the blind community.

Paul wrote: "We will spend more money creating a product that's already available and sell it to a very small subgroup of people"

I mean, this is actually a well-trodden business model (one example of many: mechanical keyboards) but yes, I do agree that in the long term this will only be attractive to a small percentage of people. I suspect that ultimately the majority of consumers will not even realize that they are buying a book with AI narration.

Tassie Dave wrote: "I'll start off by saying, I'm against AI narration as a replacement for humans. It's going to happen, but I'm not happy about it. I will always buy the human version versus the AI. If given a choic..."

You make a good point here. AI narration may never be as good as human narration but if it’s much cheaper/ easier to produce then I can soon see us getting to a place where most books have an AI-narrated audio version, and only a few choice titles get a human-narrated version. The same way that only certain books get a fancy hardback edition with sprayed edges.

I found a Corpse Hole band:

Corpse Hole, Butcher's Nails live at Pink Zeppelin San Antonio, TX April 26, 2024
https://www.youtube.com/watch?v=CxryS...

AI can already imitate celebrities. If you’ve ever watched any of the videos by There I Ruined It, you know how scary good it already is. And they’ve already duplicated James Earl Jones for Darth Vader. Within 5 years we’re going to have full cast performances in audiobooks where various characters are voiced by dead celebrities. “Marilyn Monroe, Orson Wells, Bette Davis and Paul Lynde head an all-star cast performing Harry Potter!”

Black Sabbath sings "Take Me Home Country Roads" (John Denver)
https://youtu.be/JnYhrS6pF4Q

John Denver sings "War Pigs" (Black Sabbath)
https://youtu.be/L5m5ycA_QHk

FYI here's a recipe for what Tom was going to have for dinner :) Japanese Hambaagu - an episode of Dining with the Chef on NHK World TV.

https://www3.nhk.or.jp/nhkworld/en/sh...