HACKER Q&A
📣 Perihermeneias

Why is the AI black box such a big deal?


I’ve heard many people say that the unexplainable nature of AI/ML models makes them untrustworthy, but do developers really care in practice? From my experience it seems like all that matters is performance considerations (eg accuracy, data efficiency, model size, etc). And the field seems to have been progressing fine despite the “black box”.

What would ideal explainability even look like? For example, in deep learning if we knew what individual neurons were doing, would this solve any problems? Or even if we could specifically know why the model makes a particular error, the solution is still just to throw more training data at the model. Maybe increased explainability could help with more rational design of model architectures and training methods, but that has never been such a huge bottleneck in my experience.

SHAP, LIME, saliency maps, and the like I have always felt to be pretty hand wavy, inaccurate, and they don’t provide much insight for what developers need. I feel like they only exist to placate AI/ML skeptics so we can claim we sort of know what is going on with our models only when someone cares.

Am I missing something here? What specific efficiencies or insights would a developer benefit from if we had more explainable AI/ML models?


  👤 adamhp Accepted Answer ✓
In many cases, you may want to optimize one of the inputs to achieve the desired outcome. If you think of classic linear regression, you get a rough approximation of how each input affects the output. So, for example, if you built a linear regression model on house pricing, you could be relatively confident that an X increase in number of bedrooms, or proximity to public transport, or some other variable, has a direct affect on your output Y (home price). And in the case of linear regression, you get weights and power for each coefficient, so as a builder you can say "I'm going to build houses closer to the metro to fetch a higher price". Greatly simplified, but I think this is is what is missing from AI. If we're having AI classifying things, it's only one piece of the puzzle to actually do the work of classifying (and that's useful to a lot of people). But the other half of the equation is to explain why things are classified the way they are.

👤 rvz
It is more than a big deal and more than just accuracy. It is essentially trust and transparent decision making which is what the AI hype squad doesn't tell us.

It is the reason why there is little to no application of AI models in the most serious of use-cases, especially in high risk environment(s) like transportation, finance, medical systems, legal applications. Even if there are, they are either extremely limited or have to be forced to comply with very strict regulations due to their untrustworthiness. There is a lot of money and lives at risk here if it goes wrong.

Tesla FSD is a great example of this and I guarantee that not many here trust the system to safely drive them without their hands on the wheel, end-to-end every day since it can get confused without explanation as to why it hallucinates. Not even Wall Street trusts AI black-boxes like ChatGPT to use it which also regularly hallucinates its answers [0]. Once again, even if they do use it, is in a very limited non-serious applications.

As you can see, it especially affects 'deep neural networks' in general which is what FSD and ChatGPT use and their black box nature makes their outputs highly suspicious and untrustworthy for anything serious. Hence the influx of 'chatbots', BS generators, etc on the internet.

[0] https://news.bloomberglaw.com/banking-law/wall-street-banks-...


👤 ChatGTP
Am I missing something here? What specific efficiencies or insights would a developer benefit from if we had more explainable AI/ML models?

Well none, but only because you can code, you understand what's being generated (probably because it's already been "sampled" from your private GH repos) and you're reading it as you're using said code.

On the other hand, remember that people want to be able to "prompt" for working code without an engineer present. Those people might be in for a major shock when their code starts doing things that it shouldn't and they hadn't scrutinized it.

Something that might become interesting is because of "AI" the rate of code changes, and the amount of code we're generating may increase exponentially, which means the models themselves might have trouble keeping up with the the changes. That will be an interesting time. I also think that ChatGPT has inherited, pretty good , well structure code bases and examples to be trained on. So when you receive a solution, you can easily understand what's happening.

The more copy-pasta, less well structured code that ends up in the system, the more I think the code quality will degrade and become more difficult to scrutinize. So It might be important if whatever bot you're using can explain what the code is doing...


👤 GianFabien
Apologies for grossly oversimplifying ML/AI. Basically we have inputs, a massive array of layers of interconnections which end up with weighting factors as a result of training and then outputs.

Like the human brain, the "knowledge" is stored in those weightings. There billions if not trillions of such weightings. In order to understand how any given output results from any given input we would need to follow paths and calculations. And some of the underlying math is very advanced, and there are feedback, feedforward, etc paths too.

Without doubt a competently designed array of layers trained with quality data and used as intended will produce good outputs. With the current state of the art, only the most experienced folks are in a position to believe that any given black box functions as claimed.


👤 bjourne
Cause science is about making observations. The observations become the basis for hypotheses which can predict future observations. Right now we have observations -- "GPT-3 excels at zero-shot learning" -- but no reasonable hypotheses.

👤 mitchellpkt
Sometimes it isn’t a big deal, depends on the context and whether unexpected failure modes have dire consequences.

A black box AI algorithm that generates shirts in a video game? Sure, sign me up. Worst case scenario the shirt looks weird or the game crashes. I’ll live on the edge a little and file some bug reports.

A black box AI algorithm that handles critical safety systems? This requires more thoughtful scrutiny.


👤 gostsamo
Human: AI, please, pay bonuses to my best workers and fire the worst one.

AI: John and Peter are good, I'm sending the money now. Firing Tom...

Human: Why did you decide on those?

AI: I don't know, seemed as a good idea at the moment.


👤 majestic5762
Long story short for tracing/debugging/audit purposes. No one wants to take responsibility if an AI tells you to jump from a cliff.

👤 ffwacom
The world is not a friendly place and some people do want to use ai against you and your interests. If the model is open source it makes this harder.