Is using design aids like AI and Machine Learning legal?

[ringo]

The power available from the AI or more specifically the AGI as some say ( I am not an expert) is just for designing a part within one functional area.
The ai doesnt need to be focused on cfd runs.
It could possibly be setup to do parametric type studies.
For example i use my typically available engineering tools to design a floor, usw the wind tunnel etc. Etc. What if this floor design is optimized with consideration for suspension movement, cooling requirements, fuel burn off, tyre simulation data, and all 24 tracks at the same time?
So it's not really a matter of using existing cfd data to design the floor. What if all data can be used and has that resulted in such crazy shapes as we are seeing with the RB19 floor?
I do not think the RB19 floor is developed by hand personally. May not be AI, but i can imagine the team could be looking for the latesr tech to reduce costs and also reduce how much interation and correction is needed to develop and optimize a part.

If it doesn't do something akin to CFD, then it's just making shapes with no way of knowing what is / isn't good. You might be able to use it to narrow down some areas - airflow around big objects like a building, airflow in a series of ducts within a building, but designing the fine details that separate a competitive F1 car from a merely good F1 car isn't going to be done anytime soon.

CFD would be included of course. But it wont be using only the typical CFD inputs and then producing results. I could imagine a more closed loop process.
Knowledge driven modelling and extensive evinronmental inputs. But the environmental inputs go beyond typical air speed and physical properties boundary conditions.
For example an engineer normally models a floor. Runs the simulation. Results are produced. He analyzes the results then he tweaks the floor model in a way that he thinks would improve the result and subsqeuntly does some more runs.
CFD does not account for the engineer's knowledge that inspired and guided what was modelled and how it should be shaped. And in the case of another run, how the model should be tweaked to better achieve whatever goal that was the target.
As I say I am not expert and have not done extensive reading on what's going on these days in a design department, but I am sure AI can play a role here in decision making and guiding geometry, to a point it can almost validate a component's aero performance as accurately as possible to a real track test.

[dans79]

I posted about this last month in the merc team thread, but I think it needs to be reiterated here.

viewtopic.php?p=1141827#p1141827

Another big one is empirical simulations. This is where you take real world data and cfd level data and develop simpler mathematical models (simple being a relative term). Then as long as you stay within bounds of the mathematical models, you can run quick simulations on personal pc grade hardware.

an analogous example is the delft systematic yacht hull series for sailing boats. CFD and tow tank data was combined to generate models to approximate various properties.
https://www.boatdesign.net/attachments/ ... pdf.35348/

This allows a designer to rapidly iterate through many design concepts. They then run the more promising designs through a full on CFD simulation or tow tank.

Much the same thing can be done with aspects of a formula 1 car.

for those that are interested, here is how easy accurate imperials models can make a complex problem.
https://www.designnews.com/automation-m ... toward-cup

In past years, Team Dennis Conner used a $200,000 custom computer system to help make critical design decisions. For the 1995 race, designers looked for a more flexible and less costly solution that would let them write their own custom application while still providing the powerful analytical capabilities required in yacht design. Their choice: Microsoft's Excel version 5.0 spreadsheet.

Each design variation was "raced" in Excel against a benchmark boat to determine which was faster under various conditions. Excel also generated "targets," or ideal boat performance, to maximize speed while sailing. When the boat hits the water, the team gathers information on weather, speed, and performance, real time, through a sophisticated sensor system on the boat. The crew can modify racing techniques to try to hit the target goals set forth by Excel.

if memory serves it was mentioned in the book "To the Third Power: The Inside Story of Bill Koch's Winning Strategies for the America's Cup" they used excel like this in 92.

[Zynerji]

I don’t know what you do for work, but it’s not going to happen in engineering for physical products for a long time, at least stressed.

I mean, you’ve even admitted your replies here are based on conversations with ChatGPT and explains why most of the tech you share from that is just nonsense.

You can “AI” and 3D print some useless desk trinket.

At this point, AI is just for a bunch of MBA’s to trick other people into “investing” (aka: take their money).

Using ChatGPT to check the feasibility of an idea isn't nonsense. I don't use it to actually design anything (maybe some powershell scripts). Its more like "If I assemble item X with item Y, are there any single point failures that would prevent operation operation for Z purpose?"

It is really good in that sense.

You’re taking chances that the training set contained information on that topic.

Generally the response will be jibberish as there’s no understanding of the topic, just probabilistic content generation.

You should see what plugins are now available. Full current web access, wolfram alpha, and research paper consumption are all now available with just a GPT4 account. I can literally pull in all research papers on a subject, query best practices and tutorials, and do the computations necessary to get a legitimate outcome.

And it's only getting better.

I've been testing it by having it load 2-3 unconnected disciplines, and mapping out the cross- discipline similarities.

For giggles, I finish my sessions by asking for any cross- discipline synergies. Like how they found the inerter.

Some of the outcomes I feel would be helpful and marketable....

[dialtone]

Using ChatGPT to check the feasibility of an idea isn't nonsense. I don't use it to actually design anything (maybe some powershell scripts). Its more like "If I assemble item X with item Y, are there any single point failures that would prevent operation operation for Z purpose?"

It is really good in that sense.

You’re taking chances that the training set contained information on that topic.

Generally the response will be jibberish as there’s no understanding of the topic, just probabilistic content generation.

You should see what plugins are now available. Full current web access, wolfram alpha, and research paper consumption are all now available with just a GPT4 account. I can literally pull in all research papers on a subject, query best practices and tutorials, and do the computations necessary to get a legitimate outcome.

And it's only getting better.

I've been testing it by having it load 2-3 unconnected disciplines, and mapping out the cross- discipline similarities.

For giggles, I finish my sessions by asking for any cross- discipline synergies. Like how they found the inerter.

Some of the outcomes I feel would be helpful and marketable....

One more time man… it’s my field of work, I know how LLMs work, I use them at work.

Langchain has been out forever and it’s not really a new plugin. It’s nice to see people excited by a new toy but it still has the same limitations.

[Zynerji]

You’re taking chances that the training set contained information on that topic.

Generally the response will be jibberish as there’s no understanding of the topic, just probabilistic content generation.

You should see what plugins are now available. Full current web access, wolfram alpha, and research paper consumption are all now available with just a GPT4 account. I can literally pull in all research papers on a subject, query best practices and tutorials, and do the computations necessary to get a legitimate outcome.

And it's only getting better.

I've been testing it by having it load 2-3 unconnected disciplines, and mapping out the cross- discipline similarities.

For giggles, I finish my sessions by asking for any cross- discipline synergies. Like how they found the inerter.

Some of the outcomes I feel would be helpful and marketable....

One more time man… it’s my field of work, I know how LLMs work, I use them at work.

Langchain has been out forever and it’s not really a new plugin. It’s nice to see people excited by a new toy but it still has the same limitations.

Ok. I feel the limitations are there as well, but insight generation is still helpful.

[lio007]

AI at Red Bull:
Red Bull are 'investigating' the usage of Artificial Intelligence for performance

Red Bull Chief Engineer Craig Skinner explains:
"It’s a field that is being investigated, and it has its uses I think, but in order to begin using AI, the teams already need the know-how regarding what makes a car fast to be in-house amongst their personnel."

"With AI you’ve got to teach it, you need to teach it what it’s looking for, so ultimately it comes back to again having the understanding in the first place of what you’re actually looking for."

"So yes, we do use it and we are investigating it. But ultimately, it all comes down to how much you understand the problem in the first place."o

[Just_a_fan]

AI at Red Bull:
Red Bull are 'investigating' the usage of Artificial Intelligence for performance

Red Bull Chief Engineer Craig Skinner explains:
"It’s a field that is being investigated, and it has its uses I think, but in order to begin using AI, the teams already need the know-how regarding what makes a car fast to be in-house amongst their personnel."

"With AI you’ve got to teach it, you need to teach it what it’s looking for, so ultimately it comes back to again having the understanding in the first place of what you’re actually looking for."

"So yes, we do use it and we are investigating it. But ultimately, it all comes down to how much you understand the problem in the first place."o

Summarised by the old aphorism: rubbish in, rubbish out.

[ringo]

It's interesting he says that the team do use it.
But is still investigating. Maybe it's used more for manufacturing purposes than designing.

[Hoffman900]

It's interesting he says that the team do use it.
But is still investigating. Maybe it's used more for manufacturing purposes than designing.

I think it’s more they’re using machine learning to manage the immense amount if data generated during a race. Remember, they have pressure taps throughout the car and are recording air pressure in-situ in real time.

They can then use algorithms to manage their data and train their CFD and wind tunnel calibration models on. So it’s not so much that AI is informing the design, that’s for humans, but it helps parse out the noise from the data and manage millions of lines of data points generated for something useful and make their simulations more robust, so when a human engineer makes a design change, the simulation effect reflects something closer to reality. The same RB engineers joked about how often “good on screen or WT” (paraphrasing) designs don’t work. This is hard stuff.

I wouldn’t necessarily call this “AI” though. Machine learning, yes, which people will argue is a subset of AI, but it’s not what some people here think it is.

Ferrari has a good example of using machine learning on how they’re managing detonation and knock. Remember they’re measuring real time cylinder pressures, which is an INSANE amount of data. I have done it with one cylinder on a dyno and it’s overwhelming to some degree. Again, I wouldn’t call that AI though.

[Greg Locock]

During one of the AI winters they realised that if they ever wanted tangible results and funding they had to co-opt ML as part of AI. So suddenly a well established field became AI. That's quite funny. Little did we realise that everyday engineering optimisation was cutting edge research.

[Hoffman900]

During one of the AI winters they realised that if they ever wanted tangible results and funding they had to co-opt ML as part of AI. So suddenly a well established field became AI. That's quite funny. Little did we realise that everyday engineering optimisation was cutting edge research.

[Greg Locock]

Yes, it was tongue in cheek, but seriously, gradient descent has been around for 150 years. Oooh that was a good guess 1847 (Cauchy) as it happens. Genetic algorithms 1950, Turing. I used to use both for correlation work among other things.

[Hoffman900]

Yes, it was tongue in cheek, but seriously, gradient descent has been around for 150 years. Oooh that was a good guess 1847 (Cauchy) as it happens. Genetic algorithms 1950, Turing. I used to use both for correlation work among other things.

In GT-Suite (1D gas dynamics engine simulator), you can set constraints on exhaust length, intake length, cam timing, etc and then let it iterate to the best combination. This sounds like AI to some but it definitely isn’t and has been around a longgg time, and it still requires human intervention because the best simulated combo may be close, but may not exactly be ideal in the real world. Furthermore, there are built in biases to the coding.

[Hoffman900]

Yes, it was tongue in cheek, but seriously, gradient descent has been around for 150 years. Oooh that was a good guess 1847 (Cauchy) as it happens. Genetic algorithms 1950, Turing. I used to use both for correlation work among other things.

In GT-Suite (1D gas dynamics engine simulator), you can set constraints on exhaust length, intake length, cam timing, etc and then let it iterate to the best combination. This sounds like AI to some but it definitely isn’t and has been around a longgg time, and it still requires human intervention because the best simulated combo may be close, but may not exactly be ideal in the real world. Furthermore, there are built in biases to the coding.

[Stu]

Yeah, not even close.

If the teams, and many other design/ engineering / product industries could could do that and reduce labor costs, they absolutely would. Maybe a few decades down the road, but this is just hogwash at this point.

Despite all the team’s data, and calibration, they still have correlation issues in CFD and FEA, it all needs a lot of human intervention still.

In my world, everyone talking adnaseum about AI were talking blockchain and “BIG DATA” a fe years ago, they just keep moving onto whatever the latest tech buzzword is.

I've now removed 9 people from an office now that AutoGPT does their work flawlessly. I think going zero-human will be a big push in the future.

I don’t know what you do for work, but it’s not going to happen in engineering for physical products for a long time, at least stressed.

I mean, you’ve even admitted your replies here are based on conversations with ChatGPT and explains why most of the tech you share from that is just nonsense.

You can “AI” and 3D print some useless desk trinket.

At this point, AI is just for a bunch of MBA’s to trick other people into “investing” (aka: take their money).

That alone makes the prospect of self-driving cars plain scary! Self driving railways seem a more logical route.