Use of AI for RoadTest/Project applications

I think some words in your post itself gives the answer...…never!

I think the same. I don't have problems with a Roadtest application written by AI. For example, if that is for a power supply, it creates a more or less standard plan for a power supply - test load, noise, etc. I tried that using ChatGPT 3.5. Coming up with some unique ideas will probably never happen if that was not part of the learning data.

Executing the plan is a different story. If the applicant has no idea about electronics, it will come up sooner or later. Of course, AI could be used for help too, why not. If the end result is, that the applicant learned about electronics using the roadtested product and AI, why not.

However, I would hate using AI just to generate content without any meaning. Paragraphs of bloat just to make the article longer etc. Internet is full of it even with "hand-made" stuff, especially videos. 20-60 minute videos saying just one minute of interesting information is a waste of time (but good revenue generator, apparently).

What is unique is the creativity of the plan, how it is executed, applying real-life experience, solving a real problem.

I guess we are almost in a state that we could stop using the web. Our daily feed of whatever we want to read, hear and see could be generated using AI. But clever people I know seem to be looking for the opposite. Getting information from someone knowledgeable and trustworthy in the field, with some personal spice added.

dougw said:

There was a whole document design paradigm around only needing to retype one page when a serious error had to be corrected.

How many "this page intentionally left blank" insertions can you get away with ? 

dougw said:

We had one secretary who could fix all spelling mistakes, all grammar mistakes and sentence structure all while typing 60 wpm.

I've experienced control programmers who can do similar but whilst also holding a conversation with you about the functionality required for an upcoming function yet to be written.

There was a whole document design paradigm around only needing to retype one page when a serious error had to be corrected. Retyping a whole document was a major gaffe. The typing  pool had names for people who insisted on numerous major rewrites.

We had one secretary who could fix all spelling mistakes, all grammar mistakes and sentence structure all while typing 60 wpm. It was like having an AI, except it wasn't artificial.

colporteur said:

Wow, I thought those memories were lost:)

They are parked there right alongside the maths teacher's "you won't always have a calculator to hand"...

In 81 I hired a stenographer to type my thesis for college. $180 for 50K words. It was the final requirement for my degree. Every correction was another full page charge. Wow, I thought those memories were lost:)

Probably I should have developed a little bit more my question.

I'm not thinking on using it. This question came to my mind when applying for a Project, I thought it would be unfair if someone just asked the AI to write an application, and it was selected. A big part of a RoadTest or a Project is the application, so I will always think by myself. Why would I apply using an idea that it's not mine? Could I successfully complete the tasks in that case?

However, some of you explained some use cases that probably would be ok and could be a help: translating, grammar checking...

This reminds me of the '80s when at one point the use of word processors were banned for assignments as it was seen as giving an unfair advantage over handwritten submissions. ( Yet the final assignment often used to have to be typed up and most people outsourced that to a typing service. )

Anyone using a modern writing assistant though is already likely using some form of AI, as is anyone using a programming assistant.

I'm hoping that the use of AI (for purposes other than search, translate or grammar)) can be banned from Road tests.

Why ?

When I read someone's written work it's the same as when they talk to you - there is a direct connection from one human to another. I am happy to forgive other humans their foibles and eccentricities  and to be enthused by them or their interests and insights.

Such connection is not (currently) possible with a machine - and I wouldn't want it if it were.

So when AI is used there is a risk that I'm being fooled by a machine masquerading as a person - which I don't want.

We know about the issues with AI bias, lying and confusion - why would we want to let this sneaky stuff into Road Tests ?

MK

I'm not totally against the use of AI in competitions, although I think it should be clearly identified in the blogs. If it is a road test application, it should also be stated if generative AI will be used in the road test review. I won't be using it on element14 but its use will be a critically useful/necessary skill going forward, so practicing with it is not a bad idea. If it is used, I expect the judges will take that factor into account when they choose winners.