Comment by alexasmyths | Hacker News Reader

Comment by alexasmyths | Hacker News Reader

Go Montreal!

But ...

It's just another way to transfer the exceptionalism of the local talent into the shareholders of big American companies.

I know a few in the AI-crew in Montreal and some local government types.

What they fail to grasp is that we need 'companies' that sell 'products' and 'services' and have 'revenue' - not just a bunch of shiny R&D.

A few AI jobs will actually have zero impact on the local economy.

Now - if some cushy, ridiculous 'social network' were based here, it would employ thousands, and have thousands of others supporting them in other ways, they'd be paying taxes here, and forming a hub.

Unfortunately, AI may not be the best hub for an industry - as plain AI doesn't generally make for good products. It's almost inherently a 'component' type technology. It's going to go into 'everything' but the best positioned to take advantage of it are those who already have leverage, i.e. Valley companies, finance etc..

Crossing my fingers for Mtl though.

Well to me it does seem, at least, as though the recent surge in public and private AI investments (Microsoft, Google, now Facebook) would have a positive local impact for AI researchers here. I mean with enough momentum it becomes self-sustaining with positive feedback loops: researchers have multiple local opportunities, so it makes sense to stay here in Montreal long-term, companies (even though they only need AI "as a component", as you say) have strong incentives to establish AI labs here, AI grad students have local job offers and don't get told they'd better leave for better career prospects, etc.

It's a bit early to conclude concerning the long-term "impact on the local economy" as you mention, but I guess it can't hurt to shine internationally in one more respect.

EDIT: I'd add also that there are quite a few AI-centered startups, where AI is the forefront and not just a component, e.g. look at this:

https://medium.com/@shivon/the-current-state-of-machine-inte...

and startups certainly contribute to the local economy. A local ecosystem around AI could foster such startups, I guess, so again not hurting.

It's a positive development.

But I'm not sure really that 'startups' contribute to the local economy.

The average restaurant has more revenues, and employs more people than the average startup.

The real value of most startups in Montreal is almost nil.

The 'startup contribution' generally only happens when there are large, incumbent, dominant entities that are formed. This is because of mass employment, and all of this spin-offs. Particularly - 'large dominant entities' are cash-flush and spend a lot on spurious acquisitions, etc. etc..

80% of acquisitions fail, but it still makes economic sense for big companies to buy, as there is potential upside. Without cash-flush behemoths, startups have nowhere to park - either they do an IPO or stay private forever - which is fine, but it's not generally how the ecosystem works well.

Bombardier, based in Montreal - employs 66 000 people (!!!).

$16 Billion in revenue.

So, just a 1% change in their head-count = all of the startups in Montreal.

If Bombardier secured a decent set of contracts for their mid-sized Jet - that's a definite boon for Montreal.

AI 'revenues' in Montreal ... I don't think have come close to exceeding the investment the provide/feds have made in it.

I think they will break even on it, but we need companies, not tech.

Concerning the contribution of startups, I don't have much hard numbers, but my comment was motivated by this article I recalled reading:

http://affaires.lapresse.ca/economie/201611/10/01-5039742-le...

which states that Montreal startups (by their definition) employ around 10,000 people. Of course those behind that study - local "startup scene" people and governmental instances - would have some incentive to inflate the numbers, but still, it seems larger than what you assume (1% of 66000 ~= 600 - 700 jobs). It's not Bombardier, but it's not negligible either.

I'm not sure what to say about your larger point, because I'm not sure if you're advocating something in particular. Are you saying it would be better if Facebook, Google etc. didn't open labs here, e.g. to avoid competition for local AI specialists? Or maybe you're just drawing attention to that drawback. In any case, I'd just say I'd hope it's not a zero-sum game. With better opportunities, the pool enlarges because 1) people come from abroad and 2) it makes sense for local people to specialize (e.g. pursue a PhD) in that direction.

Right. When I heard that announcement it sounded a bit like déjà vu... Same thing happened in France. Yay, we've got all the big GAFA research labs! Why? Because they get a bunch of subsidies/tax credits. Effect on the economy? Meh.

It solves the first part of the problem: attracting a lot of local talent. Now you just need an incubator and/or VC money to seed a bunch of local startups.

This is in some respects what element.ai is aiming to supplement. And having spent a lot of time in the ML community, there's a strong urge from a proportion of the students to start their own companies after finishing their studies as well.

Yes. But saying "FB's AI lab is here" is massive marketing pitch. It certainly can't hurt.

Reply on news.ycombinator.com