37

I work in naval architecture, and I have wanted to be an AI engineer for the past few years. I've taken some college courses, acquired certifications and read through text books and have done my own projects on the side. I am very passionate about the subject. Naturally I have tried to find applications for this technology in my own field of work.

I have talked to my supervisor about implementing a specific idea, and about how beneficial it will be to our work. I did this by writing up a detailed (but brief) document, outlining the intuition behind the technology, how it would work, how it would be implemented and the immediate benefits. This isn't my first pitch however. In the past, many of my ideas have been turned down due to budget and cost restraints. With this in mind, I went and developed the code on my own time, and on my own PC, to increase my chances of approval. The only cost in implementing and testing the idea would be the electricity needed to run it. In other words, there would be no cost in terms of money or time taken from my work.

I ran this by my supervisor, and his reply was the same as it has always been: it will be too expensive to implement and we don't know if it will work. At this point I can only assume that budget isn't the reason for turning down my proposal. I was very clear that I already did the necessary work on my own time, and it will not cost anything. Other than pasting the code and running it.

The thing is, if I get this to work it will be a big deal, not just for our work, but also for my career advancement. It can circumvent a lot of manual labor and reduce human error. It will also provide a great boost to my resume and enable me to possibly move onto a career in AI. In a way, I feel like I'm being denied an opportunity to grow.

I'd like to propose the idea again, but at this point I feel like I'm just an annoyance. How can I better approach my supervisor and increase my chances for approval?

1
  • 1
    Comments are not for extended discussion; this conversation has been moved to chat.
    – Kilisi
    Aug 7 at 10:39
92

You are proposing to the wrong supervisor. You need to be making these proposals at a different company.

Naval architecture is by nature very conservative. It has to be. Too many lives have been lost due to mistakes in naval architecture. It has to follow existing practices.

AI, by nature in its current implementations, is something of a "black box". One sets up an engine, feed it a large data set to "train it", and then, feed it some other data and see what patterns are found between the new data and the training data. There is still a big unknown in how AI actually finds those patterns. (See the recent ACM Communications article on trying to find ways to expose the models that AI generates internally.)

I suspect that your supervisor sees the unknowns, intuitively knows that such will ruin any adoption of your proposals, and is giving you other reasons.

Ergo, if you want to work with AI, you need to find a company that will be willing to take on the risks because of the potential rewards. A naval architecture firm is not likely to be one of them.

9
  • 11
    "AI, by nature in its current implementations, is something of a 'black box'" I think this depends on details, but the greater point of interest in OP's case may be that it's categorically (probably) bleeding edge tech with costs and risks the manager doesn't want to take on
    – CCJ
    Aug 4 at 23:56
  • 4
    I do not think any engineering firm will take whatever AI outputs and begin building with it. AI still can be useful to propose improvements that would be checked out manually. A different point is that a) it may be a project that fails to produce results and b) even if you end with an AI that produces viable designs, you are in an industry that has been refining its products for a long time and perhaps the designs are so close to being optimal that there is no room for improvement.
    – SJuan76
    Aug 5 at 8:08
  • 10
    Could you possibly link to that "recent ACM Communications article on trying to find ways to expose the models that AI generates internally"? It may be obvious for you, but I've never heard of it so for me, it's not... Aug 5 at 9:59
  • 1
    @SJuan76 "I do not think any engineering firm will take whatever AI outputs and begin building with it." Many of them already have. For instance, take a look at self-driving cars.
    – nick012000
    Aug 5 at 12:54
  • 1
    @ivan_pozdeev I'm not finding that article with a quick look through the last year. I just remember reading the difficulties of trying to find the model and thinking that is an interesting challenge. The article might be on the need to have something solid for legal cases. 10.1145/3460218 is on deceiving AI and shows some of the limits.
    – David R
    Aug 5 at 16:48
89

There are 2 issues with your proposal.

Firstly, you seem to be more excited about AI than about contributing to fulfilling whatever goals your company has. Secondly, you haven't accounted for the total cost of ownership and governance of your proposed system.

The first point is somewhat covered in other answers. I will give a brief example of the second.

Say, you want to put a coffee machine near your workplace. You rally the colleagues, everyone chips in and you are excited to let your manager know that it is not going to cost the company anything but electricity. Sounds great! However, a prudent manager may ask you:

  • Who is going to clean and maintain the machine? Should the company hire a person for this, or pay for you doing it (while you are not doing the work you were hired to do)?

  • What are the operation/maintenance routines, how are they documented, how others will be trained, examined, certified?

  • Who will assure that the machine and the coffee are safe to use and who will take the legal responsibility if anyone gets sick or burned with hot coffee? Is the risk worth it for the company?

  • How do we know that the machine does not interfere with other equipment, e.g. if it is placed in a lab or a cleanroom?

  • What are the applicable regulations that we should be aware of and what internal procedures are needed for preparing and consuming hot caffeinated beverages?

  • What will happen to the machine if all of the original contributors are gone?

  • How the machine will be phased out/replaced at the end of its lifetime? Who will cover the associated costs, including disposing of equipment, letting go of maintenance/support staff, and updating organizational processes/guidelines?

  • What are the types of risks this coffee machine introduces, what is the plan for mitigating them, and does the benefits of having the machine exceed the likely costs?

Sounds almost silly, however for these exact reasons coffee machines in offices are usually maintained by a 3rd party taking up the above issues.

My bet is that your proposed AI solution is much more complex than a coffee machine with far more serious impact. Thus, your employer is careful to give a green light for it. Especially, if it is not with the current agenda.

As for getting your manager to give you a chance, what about proposing a small pilot to try out your idea in a controlled environment. This way you can demonstrate and evaluate the benefits of your idea without putting too much risk and unknown cost to your manager?

5
  • 8
    Speaking as an SRE whose whole job is maintaining systems that others have created, I strongly agree with this entire answer. Systems reliability is expensive, difficult, and in most cases, it's only noticed when it is lacking.
    – Kevin
    Aug 5 at 17:05
  • 1
    A good alaogy, but begs the question why doesn't the supervisor come straight and explain those real life factors - at which point OP could do something about tem too. No-one loses. +1.
    – Tim
    Aug 6 at 10:40
  • 5
    @Tim many of these issues are above the paygrade of lower managers. They may not know better or not care enough to look into it. Also, he/she may hesitate to bring the idea with an uncertain potential to higher managers to get the necessary approvals. Esp. if it does not drive his/her metrics up. Aug 6 at 11:24
  • @Tim, maybe the manager just sees this as a "black box" that only the OP would be able to maintain, and thus would lock-in the company to the OP. That may be a risk in itself. Aug 7 at 17:47
  • @Tim: I think this answer very much in line with the boss's explanation ("it will be too expensive to implement and we don't know if it will work"). This answer does a better job explaining it than the boss did, but then, this is a lot for the boss to have to come up with on the fly while the OP in his office waiting for a reply . . .
    – ruakh
    Aug 8 at 3:24
9

I'd like to propose the idea again, however at this point I feel like I'm just an annoyance. How can I better approach my supervisor and increase my chances for approval?

Do the cost and budget analysis and present that as part of your proposal.

You mentioned:

At this point I can only assume that budget isn't the reason for turning down my proposal.

How can you assume this? You have stated that your previous denials were "due to budget and cost restraints" and your last denial was "it will be too expensive to implement". There is a clear pattern that your denials all involve issues of money.

If you have not done so already, I would ask your boss to lay out in detail why your proposal would not fit the budget. Once you have this information, you can break down all aspects of your proposal and get a better idea for how much things would actually cost. You may think that the only cost is electricity, but normally there are many more costs to any project than just electricity and coding.

3
  • 6
    @JoeStrazzere OP shouldn't expect it, but some bosses are more open and transparent than others and would not hesitate to at least provide a brief breakdown.
    – sf02
    Aug 4 at 14:41
  • 1
    There's also the very real chance that a new technology powered by AI just won't work. AI doesn't solve all problems well, and can fail spectacularly. Given a choice between that and the costs of failure vs a known and working mechanism, they may not be making the wrong call by saying no. Aug 5 at 4:22
  • 1
    @JoeStrazzere deleting comments that someone has already replied to is very frustrating for other users. I don't think your deleted comment had anything bad in it that would warrant a deletion.
    – rooby
    Aug 5 at 23:09
8

"How can I better approach my supervisor and increase my chances for approval?"

Well, in my field, you would go to a software architect and bounce ideas off of them to get feedback. The architect is a high level person who can basically design complex systems in their sleep. They usually aren't supervisors (at large companies) but can be at very small ones. Anyhow, if they like your idea they would become a champion of the idea and they have the ear of management. I once had a project that took maybe 4 months of full time effort approved at a company making hundreds of millions per year, just because I had maybe 2 casual talks with an architect about my idea. Maybe it takes a little more in your field, but regardless, find the RIGHT PERSON to talk to. And don't go into it assuming that you know everything. Maybe if you ask the right questions you'll learn something important about how to get to the next steps.

Your other option is to take your idea to another company that is eager to put lots of money into new ideas and that has a track record of paying employees for inventing things. I guess one problem there is that your current company may claim that they own your invention even tho you did it on your own time you've given them enough info that they might claim otherwise. Just a guess.

2
  • This. Many companies also have an innovation or research&development department.
    – Michael
    Aug 5 at 8:30
  • 1
    +1. I'm currently an architect at a large company (supervising, though) and this is exactly our job. To decide what software is best for running the enterprise. It works. I've advised to buy-out small startups, to reject big vendors, and the board does listen. This kind of proposal is well above a line manager's paygrade (I've worked in this exact field), well below the C-suite, but enterprise architects are the bridge, the experts, and the people to talk to.
    – ZOMVID-21
    Aug 8 at 20:30
7

Short version : You need to learn about businesses and how money works in them. Businesses are there to make money, not advance your technical career.

I did this by writing up a detailed (but brief) document, outlining the intuition behind the technology, how it would work, how it would be implemented and the immediate benefits.

Intuition is not something a company would implement a business plan on, much less invest resources into.

Intutition is also not a good enough reason to go against the engineer's golden rule : if it ain't broke, don't fix it.

This isn't my first pitch however. In the past many of my ideas have been turned down due to budget and cost restraints.

This is a common pattern with engineers' ideas because engineers typically fail to include costs outside of initial development. The difference between getting something from a prototype/idea to a practical implementation that you can sell to customers without constant fear of disaster is enormous.

I feel you don't fully understand the issues and probably need to learn about real product development and implementation and maintenance costs in business.

For example as someone who has been a software engineer I can tell you that no-one has ever written completely bug-free code or code that didn't require ongoing commitment to maintenance and development.

With this in mind, I went and developed the code on my own time, and on my own PC to increase my chances of approval. The only cost in implementing and testing the idea would be the electricity needed to run it. In other words, there would be no cost in terms of money or time taken from my work.

Which is wrong. You don't really understand business costs and how they relate to what you do. You have spent a lot of time learning technical things but it seems not much learning about business costs and profit and loss. Businesses operate on profit and don't on loss. Forget technology for a while and start learning about business.

Learning about business finance would be a better investment in your career than anything else, IMO.

I ran this by my supervisor, and his reply was the same it always is, it will be too expensive to implement and we don't know if it will work.

Which means you have to show this is wrong or that your supervisor is right !

Again, this means you need to learn about the economics of business, not just your technical specialty. Engineers generally are very poor at talking about the business/money side of business. Managers generally only talk about money and you need to bridge that gap.

At this point I can only assume that budget isn't the reason for turning down my proposal.

You'd be wrong. It's your failure to look at the bottom line for the company - money - in all it's gory details that is the problem.

I was very clear that I already did the necessary work on my own time, and it will not cost anything. Other than pasting the code and running it.

Wow. I would never allow that be done to any system I had responsibility for. You may have absolute faith in your ability, but a business being so reckless as to trust you so completely with a risky new venture would be run by gamblers.

The thing is, if I get this to work, it will be a big deal, not just for our work, but also for my career advancement.

Your employer is not there to advance your career. That has absolutely no relevance to your bosses or employer.

It can circumvent a lot of manual labor and reduce human error.

If I has a penny for every time I have heard about a new idea that will do that I would be a wealthy man. However in the real world those ideas mostly fail to live up to their promises.

The idea that software will reduce human error is just a fantasy normally. It usually just moves the source of the error from the idiots in front of the keyboard to the idiots behind it. Have you done a proper risk analysis relating to your ideas ?

Also providing a great boost to my resume and enabling me to possibly move onto a career in AI. So in a way, I feel like I'm being denied an opportunity to grow.

You are not growing the way your company would want you to. You are not learning the Big Picture and are purely focused on technology. Senior people need to understand the Big Picture.

Your company has exactly zero responsibility to boost your resume or enable you to get a career in AI. If you want those you may need to move away from your existing company. However your ideas would likely still be rejected at a new company if you cannot provide a Big Picture view (i.e. money related) analysis of the cost-benefit-risk of your ideas.

1
  • 1
    "if it ain't broke, don't fix it." Very good point. In my own answer I gave the example of a time I got a project approved by going to an architect but I was solving an actual existing problem otherwise I don't think it would've been approved. Solving a problem also addresses the money issue.
    – HenryM
    Aug 5 at 15:55
3

tl;dr start small, prove the concept, and then build up

I'm going to start by thinking through what a manager might respond if given the proposal (based on the information you've given), and then what you could do about it.

Is the cost-benefit worth it? There are probably hidden costs

In reading the response of your supervisor, you've focused on monetary cost. But there are probably additional costs you haven't factored in. What you're asking from your supervisor is actually a huge deal, and could take years to agree and implement, even if everyone wanted it.

I work in naval architecture

Is your role as a software engineer, responsible for creating new programs to decrease human labour? If you have another role, any time you spend on this is taking away from your normal responsibilities. Any time your supervisor (and his supervisor, and legal, and accounting, etc.) spends thinking about and approving this is taking away from their normal responsibilities. So even seriously considering this idea might cost your company hours of working time.

we don't know if it will work

Your supervisor is wise here. Even if your product is well tested, there's no guarantee it will cope with everything "in the wild". A made up example, but you've tested it produces an error if the price is greater than the company's limit of $100, but the CEO has given approval for some $200 products you didn't know about.

How do you find and fix these? Do you role it out gradually, checking the first few uses in great detail (which takes time away from other activities?) Do you role it out to everyone, and just accept there will be bugs? In naval architecture, could these bugs mean loss of life, or destruction of multi-million dollar equipment? Are there legal certifications you need to get before this software can be used?

I was very clear that I already did the necessary work on my own time, and it will not cost anything. Other than pasting the code and running it.

Who owns this code? Any answer to this could put your company in tricky legal waters (although I am not a lawyer!). If you own the code (since it was done on your time), then the company may need to write a contract and "buy" it from you. You could open-source your code, but that brings its own complexities. If the company owns the code, then any time you spent writing it could count as overtime. Local laws or company policy may require this time be paid, or you may have broken some maximum overtime limit.

It can circumvent a lot of manual labor and reduce human error

How many hours of labor will it save per month? How many people does this mean your company can fire, and how does the wages saved versus severance paid work out? Alternatively, does this free employees to do something else, and how much money does that save/create? Might your company be breaking local laws, or union agreements, but suddenly changing these workers hours or responsiblities?

In terms of human error, how often do these mistakes happen, and how much does each cost? Will it save the company from a million-dollar law suit once every ten years? Does it stop over-orders of stock by 10% each month? It might be that you've perfectly solved a problem, but it isn't actually a problem anyone cares about. (For example, companies dumping waste into the ocean, because paying the fine is still cheaper than cleaning it properly).

So in a way, I feel like I'm being denied an opportunity to grow.

If you're current job isn't in AI (or software), then you're boss doesn't have a responsibility to help you grow (in AI). He only has a responsibility to help you grow at whatever it is you're employed to do.

How can I better approach my supervisor? You start small

As you can see, there are lots of questions that need to be answered before this becomes viable. You can try to answer them by writing a longer proposal, but ultimately "the proof of the pudding is in the eating".

So, my advice would be to start small. Is there a way you can use this technology to solve a problem just in your job, or your team? Is there a small daily frustration for your supervisor that you could remove? You might find something like "Can I try this on my own machine?" is more likely to get an answer "I don't mind, as long as your reports are still accurate and on time". Once you get it working for you, you can talk to your coworkers about them trying it. Gradually, you'll build up the process and begin to answer some of this questions. Once the software has "proven itself", you might find your manager is more willing to invest in it.

3

While others have provided a lot of valuable advice on this, I want to circle back on this inflection point you make in terms of cost and time.

The cost and time to develop something will always be less than the cost and time to support it.

You could put this together over a string of weekends on your own time (~4-6 hours per weekend for 10 weekends for a total of 40-60 hours). Now it's in production, and you're getting paged about it every other day for the next three months.

To make approaches in new technology and new techniques more palatable to an organization, put yourself in their shoes and realize that cost isn't just all the things you can see. It's about the other side of it, too - how to support it, how to maintain it, how to upgrade it, how to improve it, and if necessary, how to transition away from it.

Then factor in that the person or people supporting it are human and need to be relieved from work to do human things (time off, bathroom breaks, lunch, etc), or they may get a more appetizing offer from another location and go to that instead.

Ultimately, consider that your goal is to advance your career; maybe this is the hint you could use to advance your career in an organization that explicitly focuses on the kinds of things you're interested in. Your current employer did not bring you on board with the hopes that you could super-charge their lack of AI expertise.

2

You seem to be a pioneer in a town-builder company/field (as mentioned above, naval architecture is mostly a conservative, safety-first environment).

There is a large cultural and risk aversion mismatch. This can be moderated by a wise leader (as descibed in above Simon Wardley's post), you might become such a leader and precipitate the change of mindset, or you might need to leave to find pioneers that love to move fast (and customers who appreciate that - eg. for architecture, I can imagine a large institution will be mostly conservative, while it is possible to find individual clients who aren't afraid to have something new and different).

1

This situation is the very definition of "resume-driven development" - you're proposing a new approach entirely (or at least mostly) because you want to learn it, not because it solves the business problems (or your boss's) more effectively.

To restate your proposal from your manager's point of view. You would get experience at your employer's expense, qualify that experience by switching the project to a new technology only you know, and then leave the company. All the benefits go to you, while all the risks and expenses are on your employer. Your boss can't possibly approve that in good faith. A process change shouldn't happen until the final value and long-term maintenance considerations a) exceed the costs and b) are appreciably better than the current solution. This is especially true for the military.

That's not to say the new tech you're proposing can't be the right solution in the long run, or that you can't or shouldn't advance your career by learning on the job. For the moment though, learn new tech on your own time anyways, and then take a proof of concept to your boss along with a clearer explanation of the costs and value created for your employer. That will demonstrate to your boss both lower risk/cost and higher reward, making your proposal far more viable.

You must log in to answer this question.

Not the answer you're looking for? Browse other questions tagged .