วีดีโอ

To my knowledge, there's no built-in way to do this.

Actually, there is a way, but it's a bit of a hack. It's not ideal, but it can work in a pinch. There are unicode characters for most superscript and subscript characters you would want to use. See this comment here: community.cvent.com/forums/community-home/digestviewer/viewthread?MessageKey=d60af70b-7b4b-4d89-91a2-a2e7edc1eb6a&CommunityKey=693148f0-28e3-495e-a369-16642c2a4f4f&tab=digestviewer You can simply copy and paste the subscript characters into most text editors. For example, TH-cam comments don't have a way of doing subscripts and superscripts, but using copy/paste of relevant unicode characters, I can write a chemical formula for a chlorate ion: ClO₃⁻ Having the capability as a built-in part of the text editor is badly needed. But for most applications, this can be a workaround.

@@TanyaLairdCivil Ah... Unfortunate.

She's got the knowledge but this is infuriating. I feel sorry for her students.

Sure. A few thoughts. First, keep in mind, existing generative AI models are trained on existing datasets. Someone involved in original research is by definition working outside of existing knowledge. For example, a PhD student might devote their work to designing, testing, and validating a new type of connection. But an AI would not be capable of that, as that connection has never existed before. AI can only churn out remixed copies of things that already exist; it can't make anything new. Another issue of completely replacing structural engineers with AI is that every structure is unique. You can't just copy the design of a structure intended for one location and paste it in some other location. Our existing AI models rely on huge sets of training data. You feed the whole public internet into a big training model and create ChatGPT out of it. You use that huge volume of text to teach the computer how to write. But it's hard to see how you would create such a dataset with structural design. You can't just feed a bunch of structural blueprints into an AI model and expect to train a structural engineer from that. Often beams or columns are at the location they are not because of mechanics, but because of the needs of the client. A column on a set of blueprints might be where it is because of some weird reason known only to the client. Or consider foundations. Foundations are designed to site soil conditions, and the location and design of columns are affected by this. Your site may have weak soils on just a part of it, or some soil you don't want to disturb, and this is reflected in the design drawings. But all that nuance is lost on a training model that is trying to derive all of structural engineering from just a set of plans. Really, I am not optimistic of generative AI replacing structural engineers any time soon. We already have software that can automate some functions of structural engineering in a limited, well-defined concept. For example, software exists that uses evolutionary algorithms to optimize the design of parts using stress analysis. It is not remotely possible today to feed an AI an architect's blueprints and a soil report and have it spit out a set of structural drawings. There simply isn't enough training data to capture all the permutations. And even if these get better, keep in mind that the demand for engineers and architects is not a fixed quantity. For example, today in the US, only the most expensive homes are custom designed by architects. Most people buy mass produced tract homes. For most, the first step in buying a new house is not contacting an architect. Most just go buy a tract home. But what if the architecture and structural engineering could be automated enough to lower the cost of those services? What if an architect can design a house in 1/10th the time, and a structural engineer can design the structure in 1/10th the time? They could cut the price they charge accordingly. Suddenly, the amount of people who can afford such services increases. The artistically designed custom home goes from being a luxury of the wealthy to the norm everyone can afford. I look to history, and the example I think of is clothing. In the early industrial revolution, people were poor and typically only owned 2-3 outfits that they likely handmade themselves. With automation, people worried that we wouldn't need weavers anymore. In the end, the demand for weavers actually went up. Instead of owning only a handful of outfits, people now own entire walk-in closets full of clothes. Instead of wearing the same outfit for weeks at a time, many wear multiple outfits in a single day. Technology made it easier to make clothes, but that also meant people could buy more of them. That's the perspective to keep in mind in discussions about tech-related job losses. Yes, technology will automate some of the services that structural engineers currently do. But in turn, structural engineers will be able to do more jobs in a day and will be able to charge less for each job. More people will be able to afford the services of structural engineers, and we all end up better off for it. Now, in theory, sure, you could imagine a computer far better than any bot that exists today. You can imagine a truly general artificial intelligence that can communicate with clients, operate some robot drone that can perform site visits, and monitor construction progress. But to truly replace everything a structural engineer does, you would need to have software that can literally do everything a human can. And even if you have a computer that is fully as intelligent and capable as a person, I still don't think the profession is doomed. Even if you've can mass produce Lt. Commander Data, I don't think the structural engineering profession is doomed. Why? Because I would argue that a machine that is fully capable of performing every task that a human can, that can do so in a way that is indistinguishable from a person, that can operate in our society like any other person...simply IS a person. An AI that advanced likely has its own thoughts, desires, and goals. At that point, you really have to start worrying about the ethics of forcing such entities to do your bidding. In short, if your computer is so advanced that it can completely replace a structural engineer, forcing that computer to work for you is probably slavery at that point. And we really shouldn't be creating a slave race of intelligent robots. Read either a sci fi or a history book to learn how that ends. And if you need to pay the new structural engineer AI you just created, there's no reason to create that AI in the first place. You might as well just hire an actual human.