How Botnets Are Adapting In the Face of A Twitter Crackdown

In an attempt to limit the spread of fake news, noxious political stories and propaganda, Twitter has cracked down on the use of botnets which spread misinformation in bulk.

But the laws of co-evolution and machine learning being what they are, the bots (and their human designers) are learning to disguise themselves in order to evade the new rules. This appears to be but one stage in what is likely to become an endless battle for advantage. JL


Jon Porter reports in The Verge:

More sophisticated botnet tactics have become necessary after Twitter started cracking down. The company changed the rules of its platform to ban “identical or substantially similar” tweets posted from multiple accounts, or “bulk, aggressive, or very high-volume automated retweeting.” As a result a pro-Saudi botnet had to stagger activity to avoid spamming messages in a way that made their automation look obvious. The botnet also exploited well-known methods to spread its messages. Namely, latching onto popular hashtags and helping to push them to the top of Twitter’s trending topics.

Why Human Drivers Keep Rear-Ending Self-Driving Cars

Autonomous vehicle appear to be involved in more rear-end collisions than any other kind of accident. The reason may be that the AVs are programmed to drive very cautiously and so stop unexpectedly because their Lidar (radar) systems pick up potential threats that human drivers recognize but the computers do not, causing the AV to stop while the human driver behind, having dismissed the issue, continues.

The computerized cars will learn over time, but the problems may multiply in the meantime. JL


Jack Stewart reports in Wired:

Evidence suggests these vehicles drive in ways humans might not expect, and might not want them to. AVs make cars behind them more likely to hit them: driving herkily-jerkily, or stopping for no clear reason. It indicates a focus on safety: Better to stop for a fire hydrant than run down a preschooler. But part of being a good driver is behaving in a way others expect, which doesn’t include constantly stamping on the brakes. The runner-up of crash types is sideswipes, which appear to involve human drivers frustrated at getting stuck behind a slow AV, trying to overtake it, and not making it.

How Connectivity Is Driving Innovation

Scale beats originalism. JL

Greg Satell reports in Digital Tonto:

Behind any conspicuous accomplishment you will find a community of purpose made up of small, tight-knit groups that form random connections to other tight-knit groups in order to create something truly new. That’s how an idea becomes a movement. Usually, small groups have been working in relative isolation for some time, but when those groups connect and form a shared purpose, a cascade can ensue and the idea explodes onto the public consciousness. Change is always about networks, never nodes.

Could Blockchain Break Big Tech's Dominance of Artificial Intelligence?

Yes. By giving those who generate and use data more control over who has access to it.

The problem is managing the threat-opportunity trade-off between big company dominance and potential abuse based on secretiveness. JL

Nathaniel Popper reports in the New York Times:

A.I. experts are concerned that Facebook, Google and other big companies are hoarding talent in the field. The internet giants also control the troves of online data necessary to train and refine machine learning programs. Blockchain could encourage a broader distribution of the data and algorithms, allow(ing) AI networks to access stores of data without big company control of the data or the algorithms. Data moving through the system will be locked into encrypted bundles to run the data through machine learning algorithms without seeing underlying data. The goal is to decentralize access to data.

Panasonic Introduces Headgear To Help People Concentrate in Open Plan Offices

Yes, it's come to this. JL

Natashah Hitti reports in Dezeen:

Wearable blinkers (are) designed to limit your sense of sound and sight, and help you focus on what's directly in front of you. The prototype device is to keep people distraction-free when working in busy spaces or open-plan offices by blocking them off from their immediate surroundings. To cut the user's horizontal field of vision by 60% will encourage them to concentrate on the work in front of them.

23andMe CEO Says Biggest Competitor Is 'Fake Science' - On Sites Like Goop

As people become more involved in their own healthcare and seek more information, one of the biggest issues is the quality of the guidance they are being provided. JL

Eric Johnson reports in Re/code:

"We’re messing with people’s identity. It’s really important that we take that seriously. Wellness sites have evaded those same controls and are spreading bad info. Some of the things th(ey) promote don’t actually have scientific validity that my team would be able to stand behind. Think about the publicity around the anti-vaccine movement, and their site traffic.That health side, is where there’s  the most potential. People are hungry for information about how to live healthier. And people still are figuring out what to do with information and how to be in charge of themselves."

The Reason Artificial Intelligence Won't Fix Fake News

Nuance, context, subtlety, values, morality and implication, among other crucial determinations of reality in communications, are beyond current machine learning capabilities. JL


Gary Marcus and Ernest Davis comment in the New York Times :

Causal relationships are where contemporary machine learning techniques start to stumble. Most current A.I. systems are oriented around a different set of problems. Determining that someone is implying, by a kind of moral logic isn’t a simple matter of checking a claim against a database of facts. (Such) systems lack a robust mechanism for drawing inferences or a way of connecting to a body of broader knowledge. For now “A.I. cannot tell what’s true or false — this is a skill much better suited to humans.”