Computer Chips Built Just For Artificial Intelligence

From the article:

A.I., it turned out, works better with new kinds of computer chips…Today, at least 45 start-ups are working on chips that can power tasks like speech and self-driving cars…Venture capitalists invested more than $1.5 billion in chip start-ups last year, nearly doubling the investments made two years ago, according to the research firm CB Insights.

The article doesn’t mention it, but security is another reason for companies to invest in computer chips designed for A.I. work: most A.I. algorithms are iterative or recursive, and they run substantially faster on chips using speculative execution. The recent Meltdown and Spectre exploits have shown the security risk that this process creates in most modern processors, but there are ways to design a chip without that liability. Moreover, since these chips would be secure at the hardware level, attackers couldn’t bypass the security by disabling a software patch.

With the tremendous growth in A.I. technology, these new start-ups will have plenty of competition: if you can tolerate the risk, now’s the time to pick your winners.

You can read the full article here:

Big Bets on A.I. Open a New Frontier for Chip Start-Ups, Too


The microchip in the picture is a classic, used in one of the earliest embedded systems. The first person to send me an e-mail with the name of that system will win a $10 Amazon gift card.

Please e-mail your entry to Good luck!

Hadoop: A Brief Overview

If you work with technology, you’ve probably heard of Hadoop: based on a 2003 white paper, Hadoop is the technology Google used to index the Internet. Today, it is freely available through the Apache software foundation, and is used by companies from Silicon Valley superstars like Amazon and Facebook to more traditional businesses, such as Home Depot, Angie’s List, and Verizon.

These companies use Hadoop because it lets them store very large data sets, and analyze them quickly. Consider that Google must provide accurate search results within a few seconds to users all over the world, and you can see that value that capability provides.

I’ve written a brief overview of this revolutionary technology, explaining the history of Hadoop, and how its two main components (the Hadoop Distributed File System, and the MapReduce algorithm) work. After reading this, you’ll have a better understanding of the most important technology in the Big Data landscape.

You can find the overview here:

Hadoop Basics: What is It, and How Does it Work?

Meltdown and Spectre: What We Know, and What To Do

Last week, a team of security researchers revealed two critical security flaws affecting every Intel processor released in the past 20 years. Since Intel is the leading producer of microprocessors, it is extremely likely that every business, government, and other organization is affected. It also means that your personal computers, smart phones, and other “smart” devices are affected.

Both flaws exploit the processor’s ability to guess the fastest execution path through a program, such as Microsoft Word. Underneath the hood, these programs are just a list of instructions for the processor to execute. Some instructions let the program make decisions: for example, Microsoft Word asks for confirmation before closing an unsaved Document. At this point, the program will have two sets of instructions that may be executed: one if the user confirms closing the Document, and one if she doesn’t. Only one of these branches will be executed, so if one choice is more likely–for example, if users typically do want close unsaved Documents–then the processor can speed up the program by starting to execute the instructions to close the Document before the user confirms her choice. If she doesn’t confirm, the processor can “back up” and execute the other branch. This is called speculative execution, and because of it, Intel processors are able to run certain types of programs much faster than they otherwise could.

Of course, it takes thousands of instructions for a program like Microsoft Word to do anything substantial. In practice, the decisions it makes are about memory usage, or other technical matters. The user never sees any of this.

Both Meltdown and Spectre exploit speculative execution in order to circumvent restrictions on memory access. On modern computers, the operating system (e.g. Windows) runs in a protected part of memory that isn’t directly accessible by other programs. When a program wants to do something, it makes a system call to the operating system. This increases security, and also lets the operating system prioritize certain tasks, such as downloading critical updates. By circumventing this restrictions, both Meltdown and Spectre can access system resources directly. Furthermore, since the flaw is in the computer hardware itself, software restrictions (such as user permissions) are ineffective.

At this point, your best option is to make sure your computers and smart devices have installed the latest updates. Note that the security patch may cause some programs to run more slowly: this is because it effectively disables some types of speculative execution. Unfortunately, until Intel can design and release a new processor without this flaw, this is the only solution.

You can find more details about both Meltdown and Spectre (including the official white papers) at Meltdown Attack

Stay informed. Stay safe.

Artificial Intelligence Translates Chicken “Speech”; Foghorn Leghorn Announces Presidential Bid

Researchers at the University or Georgia and the Georgia Institute of Technology have successfully used machine learning algorithms to measure stress levels in small groups of broiler chickens, based on their vocalizations. You can ready the summary from Scientific American here:

Fowl Language: AI Decodes the Nuances of Chicken Speech

While this sounds like something out of science fiction, it isn’t that surprising: animal vocalizations must be able to communicate information, or else animals wouldn’t use them. Most of the time, the differences in pitch, timbre, and expression are too nuanced for humans to decode (although the article notes that many farmers, after years of experience with their flock, can detect its general mood). A machine learning algorithm, however, can identify the latent structures common to chicken “speech”, and classify vocalizations into groups with similar structures.

Of course, the structures have to actually be there to be found, but as the algorithms continue to improve–and they are improving very rapidly–this type of application will become more and more common. Who knows: in a few years, Alexa may tell you whether your five-year old really needs a third glass of water.