Ok. I cut my end of the world AI rant down from 3,000 words to 2,000 words, then 1,000, then 500. Now the forum seems to have gone mental with half the new posts being about AI integration and my rant has increased in size again.
But, I will keep that to myself for now and just ask one genuine question. Isn’t the inevitable end point of this dive into AI going to be the end of tools like Elements?
Why would AI even need an app to build a website if all the information is out there to build it totally from scratch?
No. Probably just the opposite. Mainstream media and Wall Street are both getting this wrong. If I had more investable money at the moment, I’d be betting the opposite of the market. (Before anyone asks, no, you don’t want to be investing in AI infrastructure; Nvidia for example, is thinking their data center investments will last 10 years; they’re wrong. They’ll have to completely redo them in 2 years, so there’s no payback on the money they’re putting into it right now.)
The problem is that not a lot of folk actually really understand what AI enables and what it doesn’t do. If you think that it’s going to be used for content, then you’re betting on all content getting stale. Sort of like if you required everyone writing anything first get it approved by a copy editor. This makes for non-unique, barely useful, and non-entertaining content, and ultimately productivity goes to zero just as fast as unique content.
On the other hand, used properly by someone who knows what they’re doing, AI tools are highly enabling and that person’s productivity goes up. Way up. I think you need to actually watch someone using AI the correct way in real time to have the Aha! moment that will make your opinion flip.
As for Elements, it’s a Lego set. You can have AI do a lot of simple things very quickly (“build me a red wall with 2x1 bricks that’s 8 bricks wide and 4 bricks tall”), quicker than you could do it by hand. But the creative designer is thinking about doors, windows, walls, roofs, floors, furniture, interior design, exterior design, unusual/original arrangements of same, and a whole host of other things to create something unique. While Elements has Components (just like Legos have various blocks), unlike Legos, Elements allows you to build blocks that have never existed before and lock them into the other blocks. And yes, you can build those new blocks with AI and a little imagination.
Using the Lego analogy, I guess if you tell AI to restrict itself to using Lego then it will build things with bricks, and everything else in the Lego catalogue, but if you give it free reign could it not give you the same result on screen simply by using code, ie the plastic that the bricks are made of?
I am not going to pretend that I know how things work but I do know that if I transfer Elements files to my host server then suddenly a website magically appears. All those files are just code.
Yes, you could tell AI to build and publish a complete website. But think about what you would have to do to change any small bit of that site. If it was built with AI, you would have to go into the generated code, find what you wanted, then change it yourself; or you will be forever dependent on AI to make changes. With AI integrated into Elements, you get a WYSIWYG Elements project that any non-coder can manage.
Isn’t that exactly where we are heading? Rather than open Elements and do a couple of edits, save it, publish and back up I can just tell my AI to do X and Y and tell me when it is finished. I won’t need to maintain any software or know anything about websites at all, just keep throwing tokens at my AI.
To be honest I am more worried about the implications of this kind of thing reported on CNN today us-military-ai-false-intelligence but, IMHO, it is all part and parcel of the current headlong dive into AI.
That’s exactly what most of us are doing. In my project plan is a directive to use the built-in Components within Elements. That leads to faster design using the same Components than I can do by hand.
On occasion I’ve seen the AI interpret that in a way that allowed it build a Custom Component, which I’m usually fine with.
When I want to try something completely different, I’ll have it build out a custom Lego, but generally I restrict that with some boundaries (e.g. Javascript, which I can understand and interpret and perhaps even tweak myself).
Nothing wrong with that. We’ve been doing relatives of that for a long, long time now.
I started hard coding every byte, bit by bit. I evolved to assembly language, which is an abstraction layer above that. Today we have many multiple software layers sitting above the hardware. Elements sits on Apple’s Foundations, for instance. So I have no problem with sitting something on top of Elements, as long as it has good performance and proven functionality.
Each layer isolates me from more and more difficult and complex issues I’d have to address. Those become someone else’s problems.
Bottom line: I’m typing something at my keyboard while using my trackpad. Whether that’s an AI prompt, an Elements command, an macOS command, a SwiftUI call, or something in even deeper layers isn’t really the point. The point is actually: if the new upper layer reduces my workload, I’ll use it. If the new upper layer doesn’t reduce my workload or increases it, I won’t use it.
Sort of off-topic for this forum, but I think still deserving of a response.
Throughout history, as science has unveiled some new tech we’ve had this same discussion (i.e. “it can be used for bad…”). We have a huge pile of literature surrounding that with the nuclear bomb. We’ll have another huge pile of it surrounding AI.
But that gets to usage, monitoring, government, and regulation, not the technology itself.
I guess my point is that if AI can do the hard coding of every byte for you then that is all you need, you don’t need any of the layers above it.
I will continue to use Elements with the limited skills that I have that get better incrementally whenever I decide to add something new to my sites. I will continue to do my utmost not to have anything to do with AI because the future with it scares me and I don’t want to contribute to that future. Unfortunately my Luddite approach to this (and to many things like social media, smart watches and using my mobile phone to make payments) will get overtaken and at some point I will have no choice.
I have been around here since V3 of Rapidweaver, now we are at V3 of Elements. Sadly, I just see an inevitability around all this so cannot see Elements having a life as long as Rapidweaver. This will be a massive shame for everyone who has ever been involved with RMS but it seems to be the future for many businesses and many people around the planet. Progress.
We see AI an assistant, not a replacement for human design skills. When we built AI into Elements, we implemented it in such a way that it’s completely optional, if you remove the AI panel you’d never know there was any AI in Elements. We feel this is very important as we know not everyone wants to use it.
My personal hope is that Elements will be around for the next 10 years and beyond!
Yep, I definitely get that for the in-built AI and none of my ranting is in any way aimed at you or the RMS crew or the addition of AI to Elements. It is the impact of the big AI that sits outside Elements that I am thinking of.
Well, maybe. I just saw the first report of someone vibe coding an OS from scratch (though even that wouldn’t have been byte by byte; it sat on some sort of underlying layer).
The reason why abstraction layers exist is because doing it byte by byte, or even at the assembly language level, is a massive process. You have no help, essentially, and you have to build all the structures and functions that we rely upon these days to get things done faster. When you build at the bottom level, you’re on your own for error checking and more.
No doubt many are scared by new technologies. I’ve watched plenty of people decide they wanted to get off the tech wagon at various points. The problem with that is you’re essentially saying no to any progress at that point, so you’d better be 100% satisfied with the way things are and not bothered by the fact that others start being able to do things you can’t do. If you were a logger in the late 1800’s, you watched your career go kaput when coal and oil took over as fuel sources.
Rapidweaver had what, a 13-year life so far? (And it’s not gone away, it’ll live longer.) Elements will live a lot longer than you might think. Moreover, I ask myself would I want to start a new site today using Rapidweaver or Elements. Hands down, Elements. Thus, as a Realmac customer they’ve done the right thing by me. Indeed, I’m pretty sure they’ve gotten way more of money with Elements than they ever did with Rapidweaver. That wouldn’t have happened if they hadn’t moved the bar significantly.
Mainstream media (and Wall Street) whipped themselves into a frenzy about software dying off because of AI. They aren’t seeing reality. The reality is that *every * developer I know of, small to big, is seeing ways that they can now increase the revenue, not go away. Meanwhile, as I’ve pointed out before, the big AI companies have made an impossible bargain. Nvideo, for instance, is claiming their infrastructure investment in server farms will have a 10-year life and Wall Street has bought that line. I know for a fact that the big servers being built today have at most a 2-4 year lifespan. You can’t ignore Moore’s Law or all the other lessons that we learned in Silicon Valley over the years. Hardware investment rarely lasts five years, best case. The advances are happening fast enough I’ll bet that the server farm being built at the edge of my town will be obsolete in 2 years.
They are building one down here in NZ, local residents did not get an opportunity to quash it because the activities were assessed at the time as having no more than minor environmental effects.
It is intended that it will be the largest user of electricity in the country and will have 84 2.5MW diesel generators on standby for the time when there is a shortage of renewable energy, which will be every year due to the limited storage capacity of our hydro lakes. They also have consent to use over 2 billion litres of water per year for cooling. Looks like no water for the locals, including the many farmers, and no power in winter for the rest of us, all to feed the AI train. I question whether it is worth it, particularly in light of Thom’s comment about obsolescence above.
I first used it in 2005 so a bit older than you think.
What’s happening is a typical Silicon Valley “scramble to scale” race. Current major contestants: Anthropic, ChatGPT, Google, Grok. The simplest way to scramble is brute force large data centers and hook the customer on “first one’s free.” One everyone is relying on the big data centers and hooked on their “free taste,” then enshitification starts, just as Cory Doctorow documents.
This is one reason why I’m encouraging people to run LLMs locally. Yes, you’ll need a top end Mac with 32GB+ (preferably +) to do so effectively, but in the long run, a 64GB M5 Pro or Max is less an investment than you’d probably end up making via the Big AI in Datacenters (acronym: BAD). Yes, it may run a little more slowly. But it’s also more private.
Another possibility is to buy a dedicated AI machine for your desktop (Nvidia makes one that runs a bit over US$5000). A bit more complicated to set up and maintain, but also very effective.
This, by the way, is why Apple is sitting so pretty with AI: on-device AI may take more on-device resources, but it scales better, is better integrated, and doesn’t really force the customer into configuration fits.
Elements as a tool is finished 
thanx to Thom - yes still exploring what my M4 pro max can do…