Hacker News Reader

Comment by TuringNYC | Hacker News Reader

Here is my annoyance with AI Hype: people seeking extra tailwinds pitch their startups as "ML companies" even with the most tangential usage of ML. It drowns out real ML companies. Most people cannot tell the difference.

At a hackathon recently, someone pitched their app as an "ML-driven app" though the only ML in there was some 1-line language translation feature they were consuming off Watson REST services for a tangential feature on their app.

Meanwhile, my submission actually used a self-trained CNN on a custom dataset using TensorFlow and changes to start/end layers on the NN. The image classification features were the core of the app and it wasn't something that was just a wrapper over an Off-the-shelf API. We actually tried multiple networks and went thru the trouble of parameterizing everything.

At the end, I wonder how many judges actually understood the difference in effort/value to the two attempts at ML.

View on HN

Difficulty level 1: The only ML in there was some 1-line language translation feature they were consuming off Watson REST services for a tangential feature on their app

Difficulty level 2: Self-trained CNN on a custom dataset using TensorFlow and changes to start/end layers on the NN

Difficulty level 3: Custom rolled 12 layer CNN trained with with novel hand labeled data.

How many people to you think know the difference between these and the fact that each step is a magnitude or more harder than the last? That's not even getting into trying to apply research grade stuff.

Why should people be able to tell the difference vs the outcome of those products, unless someone (investors?) is specifically just buying into the AI hype itself? (in which case, this is nothing different from when everything was "social")

I'm not saying people should be able to tell the difference, nor am I discounting the value of the final outcome. Just that pitching an app as ML-driven, when it is barely 1% ML-driven is deceptive and mis-informs the public.

Consider taking the side mirror from a Ferrari, putting it on a Ford Escort, and then pitching the Ford Escort as a race-car. The Ford Escort may be a great family car, but it is not a sports car. People who don't know the difference might come to think of it as a sports car, and might even question the wisdom of spending money on a Ferrari.

I get (and perhaps even agree) to the sentiment, logically at least, but realistically all marketing is deceptive. It's pragmatically more useful to understand the how's and why's of that and play along.

If that 1% is essential to the purpose of the app, it seems reasonable no matter what % of the code of the app it represents.

Going along with my racecar example...the side view mirrors are essential...so does that make the Ford Escort a racecar?

The analogy doesn't work, because "racecar" isn't a technology used in the side mirrors.

Differentiation.

Most off the shelf ML tools are not that powerful - and they don't make a difference in how you differentiate as a business. If you have a ML team and know how to build models from scratch with novel data, then you can iterate on your model and expand, something you can't do if you don't own the model.

You should show your differentiation by differentiating. But in many cases the ML differentiation isn't the same as differentiating the value prop of the product.

I got tensorflow to analyze blutooth signals off a wireless controller.

talk about dreamland autoaim

you know, something like thumbs as a correlation

Effort doesn't have much relationship with value. Nor does the complexity of the core tech. People don't pay for implementation details; geek points don't mean prizes.

Effort where it maximizes return, leveraging existing assets where possible, to solve problems for real customers, is what's valuable.

Newly built tech is a rapidly depreciating asset; the depreciation comes not only from the progressing state of the art elsewhere, but also in maintenance costs and lack of network effects. Open sourcing to offset the depreciation is risky too.

I'd say you can build real "ML-driven apps"using only APIs - I disagree that you have to build it all from scratch

Your focus on "effort" and "real" sounds a bit exclusive at the behest of your own job security.

The ML boom (specifically DL) is so interesting because of the applications. Computers recognising images better than humans for instance.

Surely outside of academia its a means to an end like anything else. If Team A wires together some high level Keras components with 30 lines of code, and does something really cool with it, is it less valuable?

If said team uses Keras as just a black-box, I would argue that it is. Not having intimate knowledge of the complex algorithms that you employ to build your services is not exactly an ideal situation to be in. If things doesn't work as expected, you can't just look it up on Stack Overflow, for instance.

> even with the most tangential usage of ML

What, basic linear regression is not cutting edge enough for you?

hackathons prizes are bullshit, didn't you know

I wasn't at the hackathon for the prize, but to find enthusiastic developers to recruit for employment, those that aren't usually found through traditional searches.

Reply on news.ycombinator.com