It's really quite interesting how A16z is playing this. I've been following the types of content that they release -- and I think their vision is that the a16z brand can almost function as a consultancy with (not only) direct channels to the enterprise, but also deep technical knowledge of their problems. In the old world consultancies most of the money went to the partners -- but top engineers didn't get to rake in the profits.
In the modern world, top engineers can band together, raise VC funding, build some stupid app and get acqui-hired for 5-10x the salary. Huge discrepancies in comp.
The natural progression is that VC funds build channels and in-house expertise on technical problems in enterprise. Top engineers raise funding and are guided by partners towards solving these problems.
The new model is not that enterprises pay consultancies to solve problems, but instead, they form long standing trust based relationships with VC's who then fund companies that solve their problems (and profit when the companies profit). A big part of making this differentiation happen is releasing content that educates leaders and implementers within such enterprises.
A16z also needs to compete with the increasing growth of in-house VC firms such as GE Ventures, Verizon Ventures, Microsoft M12, The Alexa Fund (Amazon) etc. that are trying to incubate startups these companies have already identified as adjacent to their business/needed. Apparently today 75 of the fortune 100 have a dedicated corporate VC team [1].
The Alexa Fund is a good example - Amazon wants to create an ecosystem of innovation and development around Alexa voice technology, perhaps one of the clearer examples of AI today, and is shoveling cash into the space.
Big enterprises have brought the VC model in-house, closer to R&D and the problems they need to solve because most of these companies are flush with cash and facing a declining number of good ROI bets coming out of their actual R&D departments.
To stay relevant A16z needs to pick the SMB fruit that doesn't have access to its own in-house VC biz.
Notice how the document is aimed at people who own their own biz - "What can you do with AI?" "Applying AI to your business" etc.
They're hoping to stumble on SMB teams/problems that can solve a SMB problem quicker than a solution a huge enterprise can incubate in-house, then either scale it up to a late-round/pre IPO private company or sell it off to a large enterprise for the exit.
Agreed that part of that is just asking the question - have you thought about what this new technology could do for your SMB? As well as being a thought-leader in the space and putting out some educational docs so that people know A16z has a good handle on the latest AI craze and a framework for monetizing it.
Another proof that strategy consultancies producing recommendation reports and advices and not getting involved in the implementation add no value versus the new model as you described above.
> We've met with hundreds of Fortune 500 / Global 2000 companies, startups, and government policy makers asking: "How do I get started with artificial intelligence?" and "What can I do with AI in my own product or company?"
You're definitely never in a concerning part of a hype cycle when you have a technology in search of a problem. How many of these organizations just needed someone who could write a SQL query?
I generally agree that SQL could do alot of what these companies need, and not AI/ML. That said, there is definitely proper use cases for AI/ML and SQL cannot address them all. Furthermore, employing AI/ML may not just be to solve whichever problem they are trying to solve, but it may also be used to impress investors and stakeholders through the use of buzzwords; AKA, using AI/ML may be out of FOMO, used not only to address a real business problem but also to show stakeholders that they are keeping up with trends.
Sure! My position is definitely not "ML is useless"! I've written plenty of ML and particularly think the current focus on DL is missing a lot of opportunities for augmented human intelligence (and risk mitigation of systematized bias) with explainable models.
As for showing stakeholders they're keeping up with trends: yeah I'd definitely categorize that as regrettable :-)
Well, one reason to "keep up with trends" is because if your direct competitor does use ML to unveil some market/product/business opportunity or optimization, and you (executive/CIO/CEO) weren't at the least looking into the technology, heads are going to roll.
It looks like this was launched in May 2017. There's more context in the post below:
> We’ve met with hundreds of Fortune 500/ Global 2000 companies, startups, and government agencies asking: “How do I get started with artificial intelligence?” and “What can I do with AI in my own product or company?”
> While there are many excellent tutorials out there that show how to use TensorFlow or the beautiful math behind neural network training, we couldn’t find a broad overview — a “Chapter 0”, if you will — for product managers, line of business leaders, strategists, policymakers, non-AI developers to read first before moving on to more technical materials. So building on our popular primer on artificial intelligence, today we’ve launched a microsite to help newcomers — both non-technical and technical — begin exploring what’s possible with AI. The site is designed as a resource for anyone asking the two questions above, complete with examples and sample code to help get started; no computer science degree required! Ultimately, it’s aimed at people who aren’t only studying AI in universities or labs and just want to get their hands and heads around it as they explore options for their own companies.
I appreciate this article; I'm impressed when intelligent people can take complex things and explain them in laymen's terms.
I buy the idea that AI will be like RDBMS.
Except ...
RDMBS is tangible, straight forward. Easily applicable.
AI is indirect, soft.
So while I agree AI will find it's way into most things - and - will be a critical feature of some things (i.e. it will enable self driving cars) ... I still think it's over hyped.
It's a new and interesting field that is just too vague and 'non-parameterizeable' to provide value in so many ways.
Remember 'Big Data' - it was mostly an optimization. Most businesses simply don't depend on data in such quantities, and when they can make use of it, it's often just a tweak to their business, not a deep strategic insight.
If we see an explosion in GPU type computing, wherein AI experiments are able to grow maximally, perhaps we can dream a little bigger ...
But in the meantime 'there be a lot of hype' around this subject.
I think it is soft to the extent that you need an expert to define precisely the bounds of your problem and the AI solution. However, consider the following situation today: you have millions of images that are fairly related (perhaps on a real estate listing site) and tags. You can now straightforwardly build a product for predicting hashtags and describing the photos without knowing hardly anything about AI. There are plenty of products to help engineers do this already, one of which is Google cloud's ML products.
Now imagine that at some point a lot of the AI problems get to a point where they can figure out the correct training procedures automatically. People are working on this with varying amounts of progress but I think the future looks like we'll be able to do enough of this to sell/open source something as well-defined as a SQL database.
I wonder if they made this because they're struggling to get enough high quality AI startups coming their way. A decade ago you wouldn't really see VCs do this kind of stuff.
Interesting that I always used Andreessen's: Why software is eating the world? [1] as a temper to AI hype. You can replace "blockchain" with "distributed database" or "AI" with "software" and look at the result for if the use of AI or blockchain was just buzzword lingo or actually semantically relevant.
But now they have a chapter: "Giving Your Software AI Superpowers", which breaks this technique hard.
AI's history, to me, starts with Operational Research:
> Employing techniques from other mathematical sciences, such as mathematical modeling, statistical analysis, and mathematical optimization, operations research arrives at optimal or near-optimal solutions to complex decision-making problems. Because of its emphasis on human-technology interaction and because of its focus on practical applications, operations research has overlap with other disciplines, notably industrial engineering and operations management, and draws on psychology and organization science. Operations research is often concerned with determining the maximum (of profit, performance, or yield) or minimum (of loss, risk, or cost) of some real-world objective. Originating in military efforts before World War II, its techniques have grown to concern problems in a variety of industries.
Later authorities were just (in part) rebranding OR for Darpa/Iarpa grant money.
I like this executive summary though: It is good reading for managers and CEO's who may not be familiar with AI and its possibilities. For practitioners it is mostly fluff though and any intro course will give a better overview. A real playbook has yet to be written.
This seems to be, to quote, "a pre-tutorial -- a Chapter 0". I was expecting an actual playbook, as in strategies and approaches to different challenges, a list of things ML and AI can do well at this time. For me, that would be a very interesting resource.
In the modern world, top engineers can band together, raise VC funding, build some stupid app and get acqui-hired for 5-10x the salary. Huge discrepancies in comp.
The natural progression is that VC funds build channels and in-house expertise on technical problems in enterprise. Top engineers raise funding and are guided by partners towards solving these problems.
The new model is not that enterprises pay consultancies to solve problems, but instead, they form long standing trust based relationships with VC's who then fund companies that solve their problems (and profit when the companies profit). A big part of making this differentiation happen is releasing content that educates leaders and implementers within such enterprises.