Happy Sunday and welcome to the latest edition of Inside AI! For those of you who are new, I'm Rob May, CEO of Talla, and active investor in the A.I. space. If you have an early stage A.I. startup looking for investment, send me a note. Our goal in this newsletter is not to be too technical, but to give you a very wide ranging overview of what is happening in A.I. Also note that the Google Brain Team is doing another Reddit AMA. The last one was awesome, and so mark your calendar for Sept 13th to follow along and submit questions.
If you like the newsletter, I hope you will forward this to a friend so they can subscribe too.
Or if you really want to support us, go Premium and get a special Wednesday edition each week, plus extra analysis and reports. Our August report is on sales and marketing A.I. startups. Thank you to Ascent, Kylie and Pillar VC, for being sponsors. (email firstname.lastname@example.org if you want to become a sponsor.)
-- Big Idea --
The most interesting thing I read this week was Angela Bassa's article "Data Alone Isn't Ground Truth." In an age where "data is the new oil" and the competitive advantage of A.I. companies is in the data sets they have, it is important to keep in mind the limitations of data. For example, much of the recent talk about the Google Memo, bias in A.I. algorithms, and gender and tech issues, all uses data (on both sides) to prove a point. While we don't want to abandon data as a source of truth, we do need to keep it in context, particularly when look at data summarization metrics like "averages." In fact, it is useful to check out this awesome data visualization work on how different distributions can all have the same underlying metrics despite looking wildly different.
-- Must Read Links --
Growing Up With Alexa. MIT Tech Review.
This story explores something not many people are talking about at the moment - what will the impact of all these A.I. devices be on our children? If you read the recent Atlantic article "Have Smartphones Destroyed a Generation," it's a valid concern whether A.I. is going to make the next generation better off, or worse.
How To Regulate Artificial Intelligence. New York Times.
A great article by Oren Etzioni, head of Allen A.I. Institute, that proposes 3 laws for A.I.s. They are: 1) An A.I. system must be subject to all the same laws as its human operator. 2) An A.I. must always disclose that it is not human. 3) An A.I. cannot keep confidential information without approval. Go read the whole thing.
AI Writes Yelp Reviews That Pass For The Real Thing. Engadget.
This is interesting because the ability to replicate human actions like this online will fundamentally undermine a lot of the trust we have in place because of them. In fact, A.I. made lead to an arms race around the concept of trust because of tools like this.
Transformer: A Novel Neural Network Architecture. Google Brain Blog.
Google, which is already leading in neural machine translation techniques, is back with some ideas on how to make it even better with a new architecture. The gist of it is that, using a mechanism called "attention," the neural net has more context around each word.
Documents Detail Deepmind's Plans for A.I. Research on NHS Data. Techcrunch.
This is interesting because it discusses an issue that we will see more frequently in the future - data privacy, government data that is highly personally identifiable, and A.I. research. The benefits are large, but so are the risks for misuse.
-- Industry Links --
Awesome talk given at Google about the path to General Artificial Intelligence. Google Docs.
How YouTube perfected the feed. Verge
Vladimir Putin Says that whoever wins A.I. will rule the world. Washington Post. (note Mark Cuban's tweet in response)
A Chart of which jobs are most likely to be automated away. Marketwatch
IBM Is teaching A.I. to act more like the human brain. Engadget.
Turns out algorithms are racist. New Republic
Generalizing from few examples with meta-learning. Medium
What are algorithms? Though capable of great feats, they are simply lists of instructions. Economist
Background removal with deep learning. Medium
Do we need a speedometer for Artificial Intelligence? Wired
Deep Learning is not the AI future. KD Nuggets
Google and Microsoft can use AI to extract many more ad dollars from our clicks. Wired
Designing a deep learning project. Medium
The autonomous pizza delivery experiment that isn’t autonomous at all. Technology Review
Robot caregivers are saving the elderly from lives of loneliness. Engadget
If you think AI doesn’t concern you, think again. Medium
"Cortana, open Alexa," Amazon says. And Microsoft agrees. NY Times
My journey using machine learning to find porn star lookalikes. Medium
From hyperbole to real talk, law is finally ready for AI. Ross Intelligence
Domino’s really wants robots to deliver your pizza. Quartz
Salesforce is using AI to democratize SQL so anyone can query databases in natural language. Tech Crunch
How Artificial Intelligence will make digital humans Hollywood's new stars. Hollywood Reporter
-- Research Links --
Is romantic desire predictable? Machine learning applied to initial romantic attraction. National Institute of Health
Sales Forecast for Ecommerce Using CNNs. Link.
Automated Testing of Deep Neural Network Driven Autonomous Cars. Link.
--New A.I. Products--
Panasonic is putting A.I. in its new speaker.
Conversica is launching new A.I. assistant capabilities.
Kriya AI is introducing Leo to bring startups from idea to product phase.
-- Commentary --
I read the original bitcoin paper, and some other early related papers, in late 2014 and early 2015 before I started Talla. I have to confess that while I thought blockchains were cool, I didn't realize all the implications of what they could do. But I looked at a lot of blockchain related angel investments and through those companies started to learn and understand the power of where all of this can go. If you are new to blockchains, you should read this post about how they change the economics of where value accrues in the technology stack.
While I plan to write more about blockchains in future newsletters, I want to highlight 3 reasons blockchains are important for A.I. just to whet your appetite and get you thinking about the possibilities.
1. Blockchains will allow autonomous agents to trust each other. In business, when two people transact, we look for trust. This is why we have reviews of small businesses. It's why we have the Better Business Bureau. It's why we have LinkedIn recommendations. Much of what we do as humans is about making it easy to understand someone else's reputation and thus, whether or not we can trust them. As we move to autonomous agents doing work on our behalf, how do they establish trust with each other? How do you know a bot isn't spoofed? How do you know an A.I. has the authority to do the thing it wants to do? Blockchains can solve this problem by providing decentralized trust. It is built into their design.
2. Blockchains can allow DAOs. What is a DAO? A distributed autonomous organization. Imagine a time when, to start a company, you don't hire any employees. You just write out the tasks that need to be done, and program "smart contracts" for each task on the blockchain. The DAO will have an objective, will poll APIs for information, and will execute simple tasks within a business all automatically. In fact, maybe we will try at some point to do a simple DAO here, through the newsletter, so you can follow along. I'll think about the right use cases.
3. Blockchains will reinvigorate distributed systems work. Much the way that the success of neural networks reinvigorated work in other areas of A.I. like genetic algorithms and probabilistic programming, blockchains and their decentralized nature will cause a rise in non-blockchain related decentralized systems. You can imagine a series of distributed nodes that don't store or record anything (thus making them not a blockchain) but simply providing a vote on someting, or a piece of a distributed algorithm. It would be a good way to implement swarm like emergent behavior.
I've been spending a lot of time thinking about this stuff and so some future newsletters will discuss the business of A.I. and blockchains. But I thought it best to pique your interest and give you an overview of some reasons these technologies tie together well.
That's all for this week. Thanks again for reading. Please send me any articles you find that you think should be included in future newsletters. I can't respond to everything, but I do read it all and I appreciate the feedback.
-- ABOUT ME --
For new readers, I'm the co-founder and CEO of Talla. I'm also an active angel investor in A.I. I live in Boston, but spend significant time in the Bay Area, and New York.