But unfettered access to all things smutty, dirty and questionably filthy has created a surge in censorship tools that, in theory, use algorithms and advanced artificial intelligence programs to identify porn and weed it out. Last year, Twitter acquired Madbits, a small AI startup that, according to a Wired report, created a program that accurately identifies NSFW content 99 percent of time and alerts users to its presence. Late last month, Yahoo open-sourced its own deep learning AI porn filter and there are no doubt similar projects underway at other internet companies. Big players have been sinking big money into cleaning up the internet for decades. The trouble is, censorship is a slippery slope, and obscenity is inherently subjective.

After Microsoft announced Microsoft Bot Framework at BUILD and Facebook announced Bots for Messenger, Google is turning Google Now into a chatbot with Google Assistant. Unlike Cortana, Google Assistant suggests topics for users to interact with suggests follow-up topics to continue the conversation. It also keeps a running tally of the conversation, allowing a user to scroll back and continue it from an earlier point in the conversation. Expectedly, Google Assistant will replace Google Now for Android. According to Harry Shum, the Executive Vice President at Microsoft in charge of the new AI and Research Group, Cortana was originally designed to quickly interact with the user, then get out of the way.

How do you explain machine learning to a child? Answer by Daniel Tunkelang, data scientist, search/discovery expert, led teams at LinkedIn and Google, on Quora. I'd pick a universally accessible binary classification problem: learning which foods are yummy and which are yucky. We want to teach a computer to recognize which foods are yummy and which foods are yucky. But the computer doesn't have a mouth or any way of tasting the food.

When it proves safer-than-human, they'll hit the switch.

As artificial intelligence (AI) continues to move from the summer of hype to the fall tech conference news cycle, mass confusion has begun on what AI can be used for. From fears of SKYNET, to hopes for the computer in StarTrek and Jarvis in Iron Man, the value will come from defining the proper outcomes. AI is more than just a fad. With a market size of 100B by 2025, Constellation sees the AI subsets of machine learning, deep learning, natural language processing, and cognitive computing taking the market by storm (see Figure 1). The disruptive nature of AI comes from the speed, precision, and capacity of augmenting humanity.

If you have ever bought something from an online store, chances are the store's used your email address with wanton disregard, bombarding you with email after email about its products and sales until you reach for the sweet oblivion of unsubscription. Stores and brands do this to keep customers engaged--but they don't know how many emails are too many. Adobe previewed a tool today with the promise to help alleviate what they call "customer fatigue." By using machine learning algorithms to crunch the numbers of how often emails are opened and clicked on, marketers can see whether customers are tired of getting their emails. The algorithms will allow Adobe to actually calculate how fatigued every customer might be, and only send a certain number of emails based on that score.

The procedure used to carry out the learning process in a neural network is called the training algorithm. There are many different training algorithms, whith different characteristics and performance. The learning problem in neural networks is formulated in terms of the minimization of a loss function, f. This function is in general, composed of an error and a regularization terms. The error term evaluates how a neural network fits the data set. On the other hand, the regularization term is used to prevent overfitting, by controlling the effective complexity of the neural network.

Trying to discern drug smugglers passing through customs presents exactly the same problem as trying to discern security threats passing through our networks. Machine learning has been applied to both with varying degrees of success, but ultimately the technology reaches the same limitations. Machine learning has two basic elements: feature vectors and classification exemplars -- the data that is gathered and the corresponding classification examples. In the case of drug smugglers, we might observe number of travelers, point of origin, point of destination, number of bags, length of stay and weight of the bags. We might also flag any traveler or pair of travelers with two or more bags whose combined weight is greater than 150 pounds, whose stay is less than a week and who originated from a climate conducive to poppies.

We've already started to see effective use of emerging technologies, such as Industrial IoT (IIoT)-enabled remote condition monitoring and Big Data analytics for predictive maintenance and similar offline applications; but process engineers are interested in knowing if and how emerging technologies can be used to improve the actual production process and product quality. In a recent pilot program, Mitsui Chemicals, Inc. and NTT Communications Corporation (NTT Com), the industrial control technology (ICT) solutions and international communications business within NTT Group, have successfully created a Deep Learning technique that accurately predicts the quality of gas products during production; 20 minutes before the final product is created. As we learned in a recent press release from NTT Com, the technique is based on modeling the relationship between the different data sets sourced from raw materials feeding into the reactor; reactor conditions; and the trace gas impurities that represent gas product quality, expressed here as "X-gas." The goal of this joint project between NTT Com and Mitsui Chemicals is to improve the accuracy of detecting abnormalities in product quality to improve operational efficiencies and product quality. The two companies initiated the pilot project at one of Mitsui Chemicals' gas production plants in 2015.

Machines have never replaced humans before, and they probably aren't doing so right now, argues Noah Smith Of all the economic questions being debated today, the most frightening one is, "Will the robots take our jobs?" This nightmare scenario comes in several flavors. The extreme version is that automation simply makes human workers obsolete, just as cars made horses redundant. A less apocalyptic possibility is what economists call "skill-biased technological change" -- people who are technically savvy, mentally flexible and educated will reap greater and greater rewards, while everyone else sees their wages decline. These two scenarios might look different on paper, but the net result is largely the same -- a very big portion of humanity would be either be impoverished or reduced to living off of the government dole.