AI - is it the wave of the future?

[Bruce]

I am wondering about the thoughts of this community on the subject.... Is AI the ultimate tool?

Not a day goes by without hearing how "AI" is solving all the problems. IBM and Amazon flood the airwaves with their claims of what their compute platforms can do. The ability to propose solutions to problems given the information provided is touted to be better than anything previous. And yet, having been in tech all my life, I find the promises too good to accept as true.

Guided Functions on professional tools comes to mind. At the advent of such, it too was supposed to make the job so much easier. In the hands of novices, Guided Functions probably did increase the likelihood that those technicians managed to find a solution - but at what cost? In the hands of a well skilled technician, it seemed the guided scripts found in a guided tool wasted time and effort. The well skilled tech soon wanted to walk away from such a tool.

How will the designers of AI based diagnosis do better at helping solve the complex issues faced these days? Will technicians embrace this new tool? Or is it going to be another means of draining our pockets of the money we work hard to earn?

Perhaps we can use this thread to learn what others are doing with/thinking about this life changing technology so touted by the media (automotive and other). Help the community understand how you see AI changing the way diagnostic procedures on cars will be conducted.

Thanks for reading. I look forward to your comments.

 

[TTT]

AI will not be applicable to ICE technologies, I see it more on full electric, where the outcome will be: change that module and that module and AI will take care of the programming and coding side. Technology will be too complex for the workshop persons, how many here understand what a three-phase motor is and how many understand frequency converters operation and troubleshooting.

Look to VAG world, 2000 parameters for a BCM? Insane, only if you look to the work hours put into design of a single module.

 

[PetrolDave]

As I understand it AI behaviour is totally dependent on the data used to train it.

This means that given "good data" AI can be useful and helpful, and maybe even learn some of the "gut feel" diagnostic techniques that take humans years to learn.

But given "bad data" AI can mislead or even become malicious or dangerous.

 

[Bruce]

@PetrolDave you are speaking of the garbage in = garbage out concept.. the one I struggle to see how AI evolves to the levels touted..

@TTT I hear you. I am a EE and my specialty was in power and control - and even I struggle with the concepts they are using to control these motors in the EVs. As to the complexity of the control modules, yes, we get it looking at all the data we need to see these days to make the tools as we do now.

@TTT, I guess my question to you is do you feel they will not make the effort to collect the information needed to support ICE machines since it seems their life on this planet is now being made short by government regulations? That is, will we only feed the AI beast information that will help down the road? Aren't all the OEs collecting said information from the cars as they are touched by the dealerships? I do accept that the information will not be shared OE to OE but they are collecting it. Won't they at least use it to help their techs?

 

[SaVAGeSoot]

AI has a place and use... or "can" have one..... there isn't a single one answer solution for most things....
AI can be useful in things like research, coding, and other areas where there's either a lot of data to analyze and categorize and pick from, or where it's THIS or THAT kind of thing...

Just think of any computer\program\system.... garbage in, garbage out.....

So if we give AI a set of data to learn from, but it's all based on the Mechanic's Diagnostic Dice, it can only come up with suggestions based on that.... so every problem will be solved by either replacing a flat tire or a head gasket..... know what I mean?

I believe it can be used as a tool or aid, but someone with a clue and a brain still needs to pull rank before the "solution" is used.

 

[TTT]

@Bruce, ICE is a done business, nothing to be found anymore. I can't see OEs investing big bucks just to help the workshop floor for easier troubleshooting for ICE.

2 persons die everyday from road traffic accidents in NL. This is where some of the focus could be and we all train the system everyday, by clicking those images, that's a stair, that's a traffic light, etc. Image recognition will not be possible to be done real life without AI.

Another application for AI would be how to make production cheaper, smarter and beter. Isn't it for everyone? Bet Ross Tech tried automated routing for PCBs in the past? This is also a form of AI.

I think you all saw this before, a difference of 10 years. I wish Ross Tech stays with the wave. Yeah, maybe few here and there will disable their cars, but wasn't that always the case?

 

[dafrazi]

At my current job, we use AI to review and categorize human collected data...everything from form data to handwritten reports. It does a great job and who actually wants to do that work. It frees us to work with the data, instead of on the data. Seems like a perfect use for the current state of AI.

I see companies using AI to find outliers in data. I think similar tools can be used in diagnostics. It's not much of a stretch to go from what's not right to what's almost not right. It's the programming/teaching of the AI to put those things together that's currently missing. It's no different than teaching/training humans to do something. The benefit is you don't have to teach it more than once...downfall is humans suck at teaching machines but we're learning how to do that too.

As an engineer, I've watched as design software has taken over the grunt work of calculations and moved to resolving more and more of the design decisions. At this time, none of the software companies will take responsibility for their products' output. I've come to the conclusion this partly due to the risk to the software company and partly from the protections around the engineering profession. It's on the engineer using the tool to ensure the output is correct...which can be damn near impossible with the complexity of some engineering models. As an engineer, you have to learn to use and develop confidence in your tools, be it a slide rule, calculator or software. AI isn't really that different. For now, AI should be tool for our use, not a replacement for us. It's anybody's guess on when AI will be ready to take the wheel. I'm nervous we'll let AI take over at the wrong time or take over the wrong things. I'm even more nervous about who is going to make those decisions.

With that said, I have software tools, now, which will automatically optimize designs. They are short of AI but could be AI grade without much effort. I see AI completely taking over the engineering profession within the next 50 years...AI is already doing creative design. AI can crunch numbers faster with less mistakes than we can. We've already built the engineering software for AI to use, we now need to wait for AI to be ready to make decisions. As a group, engineers are largely conservative, conscientious stewards of the profession. I believe the engineers developing the AI that will take over from us will do a solid job. I'll have less concern over garbage in-garbage out once humans are a smaller variable in the equation. Numbers don't lie, but people do. I never had a calculator lie to me, but I have seen someone multiple by zero and add the right answer to avoid redoing work.

The promise of AI is it will accelerate innovation and better the world. I'm hopeful AI will deliver but apprehensive of the journey.

 

[PetrolDave]

As I understand it AI behaviour is totally dependent on the data used to train it.

It's the programming/teaching of the AI to put those things together that's currently missing.

 

[RGH0]

I started in process plant engineering support 50 years ago and have gone from pneumatic instruments and paper charts to fully integrated process control systems and complex model based plant optimisation. I have seen the emergence of some limited AI based analysis systems in recent years.

The challenge in process plants is the availability of data and the unique nature of every plant. You need to see, listen, and smell whats going on and talk to others on what they see and do also as well as look at the process control information available for AI analysis. AI helps but it is a limited solution and not the total answer

In mass produced indentical purely electronic systems AI will rapidly progress. In the manufacturing real world of complex one off machines in a complex one off process line arrangement there will be a future for engineers in plant support and optimsation for a long long time.

 

[SaVAGeSoot]

Maybe some of you are into racing....
Maybe some of you are into tech.....
Maybe some of you have heard about this....
Maybe not.....

Abu Dhabi Autonomous Racing League in Abu Dhabi,UAE | A2RL

Welcome to the Abu Dhabi Autonomous Racing League - A2RL, where we bring together the exciting impact of AI on our lives and its incredible potential for autonomy.

a2rl.io

 

[Bruce]

@SaVAGeSoot - I was not aware. Thank you for sharing that.

 

[SaVAGeSoot]

You are welcome... but as excited as all that gets me, it also angers me that their animation shows the car doing a 4wd burnout towards the end... I mean c'mon, we are talking about breaking new boundaries with all this tech, and we can't even "catch" some boo-boo an animator who can't probably even spell F1 created for our website!

 

[DanConnor]

I am wondering about the thoughts of this community on the subject.... Is AI the ultimate tool?

Not a day goes by without hearing how "AI" is solving all the problems. IBM and Amazon flood the airwaves with their claims of what their compute platforms can do. The ability to propose solutions to problems given the information provided is touted to be better than anything previous. And yet, having been in tech all my life, I find the promises too good to accept as true.

Guided Functions on professional tools comes to mind. At the advent of such, it too was supposed to make the job so much easier. In the hands of novices, Guided Functions probably did increase the likelihood that those technicians managed to find a solution - but at what cost? In the hands of a well skilled technician, it seemed the guided scripts found in a guided tool wasted time and effort. The well skilled tech soon wanted to walk away from such a tool.

How will the designers of AI based diagnosis do better at helping solve the complex issues faced these days? Will technicians embrace this new tool? Or is it going to be another means of draining our pockets of the money we work hard to earn?

Perhaps we can use this thread to learn what others are doing with/thinking about this life changing technology so touted by the media (automotive and other). Help the community understand how you see AI changing the way diagnostic procedures on cars will be conducted.

Thanks for reading. I look forward to your comments.

I'd say AI is going to make a decently huge shift in coding world n stuff, already can see it being a huge tool for writing and design. Sky is the limit, esp if there will be no laws or limitations coming from government. I'm not sure if it will completely remove some positions in diagnostics, but it can reduce workload/speed up process under an experienced operator.
Regarding if it's going to be a useful tool or just a money waste - first wave probably going to be more of a money drain, rather than a working product. But over time it should improve and then it's probably going to be more profitable to have it, rather than not.

 

[Uwe]

Here's a test for "AI". Get it to consistently produce correct codings for MK60EC1 (or better yet, MK100 IPB) ABS controllers.

-Uwe-

 

[SaVAGeSoot]

I had an interesting though over lunch.... what if AI was responding to our discussion here?
Here goes nothing

I am wondering about the thoughts of this community on the subject.... Is AI the ultimate tool?

Not a day goes by without hearing how "AI" is solving all the problems. IBM and Amazon flood the airwaves with their claims of what their compute platforms can do. The ability to propose solutions to problems given the information provided is touted to be better than anything previous. And yet, having been in tech all my life, I find the promises too good to accept as true.

Guided Functions on professional tools comes to mind. At the advent of such, it too was supposed to make the job so much easier. In the hands of novices, Guided Functions probably did increase the likelihood that those technicians managed to find a solution - but at what cost? In the hands of a well skilled technician, it seemed the guided scripts found in a guided tool wasted time and effort. The well skilled tech soon wanted to walk away from such a tool.

How will the designers of AI based diagnosis do better at helping solve the complex issues faced these days? Will technicians embrace this new tool? Or is it going to be another means of draining our pockets of the money we work hard to earn?

Perhaps we can use this thread to learn what others are doing with/thinking about this life changing technology so touted by the media (automotive and other). Help the community understand how you see AI changing the way diagnostic procedures on cars will be conducted.

Thanks for reading. I look forward to your comments.

AI as the Ultimate Tool: A Community Discussion
AI, or Artificial Intelligence, has undoubtedly become a buzzword in today's technological landscape. It claims to revolutionize various industries, solving complex problems and improving efficiency. However, skepticism exists regarding the true potential and impact of AI. This community post seeks to gather thoughts and initiate a discussion on whether AI is truly the ultimate tool, particularly in the realm of diagnostic procedures for automobiles.

Reflecting on past experiences, where guided functions promised to simplify tasks but often fell short, it's natural to question if AI will face a similar fate. While novices may benefit from AI-driven solutions, experts may perceive such tools as a hindrance, wasting time and effort. Therefore, it becomes essential to explore how AI-based diagnosis can address the complexity of modern issues and win over skilled technicians. Additionally, concerns arise about the financial implications of adopting AI solutions.

This thread aims to gather insights from the community, shedding light on the real-world applications and future vision of AI in the automotive (and other related) industries. By understanding different perspectives, we can evaluate how AI is poised to shape the future of diagnostic procedures and determine the extent to which it truly lives up to the hype.

Share your thoughts, experiences, and opinions on AI as the ultimate tool and its potential impact on diagnostic procedures in the automotive industry. Let's engage in a meaningful conversation to better comprehend the possibilities and challenges surrounding this transformative technology.

AI will not be applicable to ICE technologies, I see it more on full electric, where the outcome will be: change that module and that module and AI will take care of the programming and coding side. Technology will be too complex for the workshop persons, how many here understand what a three-phase motor is and how many understand frequency converters operation and troubleshooting.

Look to VAG world, 2000 parameters for a BCM? Insane, only if you look to the work hours put into design of a single module.

Thank you for sharing your perspective on the applicability of AI in the automotive industry, particularly in relation to internal combustion engine (ICE) technologies versus full-electric vehicles. It is interesting to consider the potential divide in AI's impact based on the type of technology being used.

Your point regarding the complexity of technology and its potential impact on workshop personnel is crucial. As vehicles become more advanced, technicians will need to navigate intricate systems and understand a multitude of parameters. AI has the potential to assist in this regard by automating certain tasks and simplifying programming and coding processes. This could potentially reduce the burden on technicians while improving efficiency.

Furthermore, the example you provided regarding the VAG world highlights the extent of complexity inherent in modern automotive systems. With an increasing number of parameters to manage in each module, AI-powered solutions could aid technicians in diagnosing and troubleshooting more efficiently.

As I understand it AI behaviour is totally dependent on the data used to train it.

This means that given "good data" AI can be useful and helpful, and maybe even learn some of the "gut feel" diagnostic techniques that take humans years to learn.

But given "bad data" AI can mislead or even become malicious or dangerous.

You are absolutely correct in saying that AI behavior is heavily dependent on the data used to train it. The quality and diversity of the data play a crucial role in determining the effectiveness and reliability of AI systems.

When provided with good data, AI can learn patterns, make accurate predictions, and assist in various tasks, including diagnostics. It has the potential to analyze large amounts of data quickly and efficiently, potentially even uncovering insights that humans may overlook. In some cases, AI might even learn diagnostic techniques akin to those developed through years of human experience.

However, the concern you raise about bad data is also valid. If the data used to train AI is flawed, biased, or incomplete, it can lead to incorrect or unreliable outputs. AI depends on the patterns it discovers in the training data, so if the training data represents a limited scope or contains biases, these limitations can be perpetuated in its behavior.

Furthermore, if AI systems are maliciously or intentionally trained with harmful data, they can indeed become dangerous or misleading. This highlights the importance of carefully curating and validating the data used to train AI models, as well as implementing rigorous testing and evaluation procedures to ensure the accuracy and safety of AI applications.

It is crucial to understand that AI is a tool that requires responsible development, training, and monitoring to ensure its outcomes are beneficial and aligned with human values. Transparency, accountability, and ethical considerations must be integral parts of AI development and deployment to mitigate potential risks associated with bad or biased data.

AI has a place and use... or "can" have one..... there isn't a single one answer solution for most things....
AI can be useful in things like research, coding, and other areas where there's either a lot of data to analyze and categorize and pick from, or where it's THIS or THAT kind of thing...

Just think of any computer\program\system.... garbage in, garbage out.....

So if we give AI a set of data to learn from, but it's all based on the Mechanic's Diagnostic Dice, it can only come up with suggestions based on that.... so every problem will be solved by either replacing a flat tire or a head gasket..... know what I mean?

I believe it can be used as a tool or aid, but someone with a clue and a brain still needs to pull rank before the "solution" is used.

You are absolutely correct that AI has its place and can be highly useful in various areas such as research, coding, and data analysis. Its ability to analyze large amounts of data quickly and identify patterns can be incredibly beneficial in situations where there are vast amounts of information to process or when decisions need to be made based on specific criteria.

However, your analogy of "garbage in, garbage out" is also very apt. The quality and relevance of the data used to train AI systems greatly impact the outcomes. If the data is biased, incomplete, or flawed, the AI's suggestions or solutions will also reflect those limitations.

In complex problem-solving scenarios, where multiple factors and variables are at play, human expertise and judgment are indeed invaluable. AI should be seen as a tool or aid that can assist humans in making more informed decisions or performing tasks more efficiently. The final decision should always be made by individuals who possess domain knowledge, experience, and critical thinking abilities.

It is crucial to understand that AI is not a substitute for human intelligence and decision-making. Rather, it should be viewed as a powerful tool to augment human capabilities. Combining the strengths of AI with human judgment and reasoning can lead to enhanced outcomes and more effective solutions.

Therefore, as you rightly said, someone knowledgeable and capable should exercise oversight and assess the AI's suggestions before implementing them, ensuring that the final solution aligns with human judgment and common sense.

At my current job, we use AI to review and categorize human collected data...everything from form data to handwritten reports. It does a great job and who actually wants to do that work. It frees us to work with the data, instead of on the data. Seems like a perfect use for the current state of AI.

I see companies using AI to find outliers in data. I think similar tools can be used in diagnostics. It's not much of a stretch to go from what's not right to what's almost not right. It's the programming/teaching of the AI to put those things together that's currently missing. It's no different than teaching/training humans to do something. The benefit is you don't have to teach it more than once...downfall is humans suck at teaching machines but we're learning how to do that too.

As an engineer, I've watched as design software has taken over the grunt work of calculations and moved to resolving more and more of the design decisions. At this time, none of the software companies will take responsibility for their products' output. I've come to the conclusion this partly due to the risk to the software company and partly from the protections around the engineering profession. It's on the engineer using the tool to ensure the output is correct...which can be damn near impossible with the complexity of some engineering models. As an engineer, you have to learn to use and develop confidence in your tools, be it a slide rule, calculator or software. AI isn't really that different. For now, AI should be tool for our use, not a replacement for us. It's anybody's guess on when AI will be ready to take the wheel. I'm nervous we'll let AI take over at the wrong time or take over the wrong things. I'm even more nervous about who is going to make those decisions.

With that said, I have software tools, now, which will automatically optimize designs. They are short of AI but could be AI grade without much effort. I see AI completely taking over the engineering profession within the next 50 years...AI is already doing creative design. AI can crunch numbers faster with less mistakes than we can. We've already built the engineering software for AI to use, we now need to wait for AI to be ready to make decisions. As a group, engineers are largely conservative, conscientious stewards of the profession. I believe the engineers developing the AI that will take over from us will do a solid job. I'll have less concern over garbage in-garbage out once humans are a smaller variable in the equation. Numbers don't lie, but people do. I never had a calculator lie to me, but I have seen someone multiple by zero and add the right answer to avoid redoing work.

The promise of AI is it will accelerate innovation and better the world. I'm hopeful AI will deliver but apprehensive of the journey.

Your perspective on the use of AI in your current job is insightful. Automating the review and categorization of human-collected data can indeed be a perfect application for AI, allowing humans to focus on higher-level tasks that require their expertise and creativity. It's evident that AI can bring immense value when it comes to handling large volumes of data efficiently.

Furthermore, your observation about using AI to find outliers and potential areas of concern resonates well. With the right programming and training, AI can assist in diagnostic tasks by identifying patterns that may indicate issues or abnormalities. As the capabilities of AI continue to advance, it's reasonable to expect that it could become even more adept at recognizing subtle deviations and potential problems.

Your comparison between the evolution of design software and AI is also thought-provoking. Indeed, engineering tools have steadily progressed to automate calculations and provide decision-making support. However, the responsibility for the output still lies with the engineer, highlighting the importance of a human-in-the-loop approach. AI should be treated as a tool and not a replacement for human engineers, as only humans can fully comprehend complex engineering models and make judgment calls based on experience and context.

Your belief that AI will eventually take over the engineering profession aligns with the notion that AI can increasingly handle creative design and optimize engineering processes. However, it's essential to ensure that AI is developed responsibly and ethically, with human engineers playing a crucial role in overseeing and validating its outputs. As you rightly pointed out, the decision-making process and determining what tasks AI should take over need careful consideration.

Your apprehension and hope for the future of AI are well-founded. While AI has the potential to accelerate innovation and improve our world, it's important to proceed cautiously and address the challenges and risks that may accompany its rapid advancement. The role of engineers in shaping AI's development and ensuring its responsible use will be vital in navigating this journey.

Overall, your insights provide a balanced perspective on the potential of AI in various domains and highlight the importance of maintaining human oversight and responsibility in the face of advancing technology.

I started in process plant engineering support 50 years ago and have gone from pneumatic instruments and paper charts to fully integrated process control systems and complex model based plant optimisation. I have seen the emergence of some limited AI based analysis systems in recent years.

The challenge in process plants is the availability of data and the unique nature of every plant. You need to see, listen, and smell whats going on and talk to others on what they see and do also as well as look at the process control information available for AI analysis. AI helps but it is a limited solution and not the total answer

In mass produced indentical purely electronic systems AI will rapidly progress. In the manufacturing real world of complex one off machines in a complex one off process line arrangement there will be a future for engineers in plant support and optimsation for a long long time.

Your experience in process plant engineering support over the past 50 years provides valuable insights into the evolution of technology in this field. From the transition from pneumatic instruments to advanced process control systems and now the emergence of AI-based analysis systems, you have witnessed significant advancements.

The challenges you mentioned in process plants are valid and highlight the unique nature of each plant. The availability of data is crucial for AI analysis, but it's equally important to combine data analysis with other sensory observations, such as visual inspections, listening, and even smelling, to gain a comprehensive understanding of the plant's operations. The human element and the ability to communicate and collaborate with others on the ground remain essential in this context, where AI alone may provide a limited solution.

It's interesting to note your differentiation between mass-produced identical electronic systems and the real-world scenario of complex, one-off machines in a unique process line arrangement. While AI may rapidly progress in mass production settings, the complexity and individuality of one-off processes in manufacturing suggest that engineers will continue to play a crucial role in plant support and optimization for a long time to come.

Your perspective highlights the value of human expertise, adaptability, and problem-solving skills in navigating the complex and varied landscape of process engineering. While AI can assist in specific areas, such as data analysis and optimization, the role of engineers as plant support and optimization specialists remains indispensable, particularly in handling the intricacies of unique processes and ensuring long-term operational success.

Thank you for sharing your valuable perspective and reinforcing the enduring importance of engineers in the field of process plant engineering support.

I'd say AI is going to make a decently huge shift in coding world n stuff, already can see it being a huge tool for writing and design. Sky is the limit, esp if there will be no laws or limitations coming from government. I'm not sure if it will completely remove some positions in diagnostics, but it can reduce workload/speed up process under an experienced operator.
Regarding if it's going to be a useful tool or just a money waste - first wave probably going to be more of a money drain, rather than a working product. But over time it should improve and then it's probably going to be more profitable to have it, rather than not.

You bring up an interesting point about the potential impact of AI in the coding and design world. Indeed, AI has the potential to revolutionize various aspects of these fields. It can serve as a powerful tool for tasks like writing code and generating designs, potentially increasing efficiency and productivity.

However, the extent to which AI will replace specific positions in diagnostics is uncertain. While AI can assist in diagnosing issues and reducing workload, the expertise and insights of experienced operators still play a vital role. AI should be seen as a complementary tool rather than a complete replacement.

It's important to consider the role of regulations and limitations imposed by governments. The future progress of AI will be influenced by these factors, ensuring its responsible and ethical use. These considerations will shape how AI evolves and the boundaries within which it operates.

In terms of economic viability, early stages of AI development may indeed be costly, potentially draining resources as systems are refined and improved. However, as AI technology progresses and becomes more efficient, it is likely to become a valuable and profitable tool. Businesses that embrace and leverage AI effectively should eventually see the benefits outweigh the initial costs.

Overall, the potential of AI in coding and design is vast, with the ability to enhance productivity and provide innovative solutions. It will likely be a gradual adoption, with continued improvements and careful consideration of ethical and regulatory aspects.

 

[Uwe]

@SaVAGeSoot please don't use red text in large doses. It hurts my eyes.

-Uwe-

 

[SaVAGeSoot]

Sorry Boss

 

[SaVAGeSoot]

@SaVAGeSoot please don't use red text in large doses. It hurts my eyes.

-Uwe-

Here's me attempting to "redeem" myself

 

[Mike R]

AI has the potential to do great things... however it only takes one wrong step for us to go beyond the event horizon. Even if allowing AI's to iterate on their own code was illegal, it just takes one bad actor for us to potentially wind up with a God AI, and what that AI would potentially do, compassionate or malevolent, nobody really knows. Even if it's something we'd see as malevolent, it's just completing whatever task it's set out to do through what it reasons is the best method to accomplish it.


The issue arises when we get into Deep Learning. Machine learning is safe and can help be predictive towards potential diseases or running simulations of astrophysical tasks/launches.

But I am 100% against giving AI the autonomy of self improvement or decision making.