atans1

Posts Tagged ‘AI’

Computers can think out-of-the-box

In Uncategorized on 05/09/2018 at 5:50 pm

So there’s no need for humans any more except to build these computers.

I was reading an article in the FT on quantum computing and read that not only can they think out of the box but they are better than humans because of the speed that they can think such tots.

Advertisements

AI is a problem for India and PeenoyLand

In India, Internet on 28/08/2018 at 7:04 am

And call centre workers everywhere.

The biggest threat to jobs might not be physical robots, but intelligent software agents that can understand our questions and speak to us, integrating seamlessly with all the other programs we use at home and at work. And call centres are particularly at risk.

BBC https://www.bbc.com/news/business-45272835

It reports Brian Manusama, an analyst at market research firm Gartner saying:

“The number one use case for applying AI is in this call centre and customer service space”

and

“At the end of 2017 about 70% of all use cases in AI were related to customer service and call centres.”

Scale of problem

Several million people are employed in call centre roles in the US and UK and hundreds of thousands more rely on such work in countries like India and the Philippines. Unless these people quickly learn new skills, they could soon be out of work.

Modified to attribute the story to BBC, not FT. Sorry.

Juz be smarter than AI

In Internet, Uncategorized on 24/08/2018 at 6:40 am

Jason Karp, who runs the long-short equity hedge fund Tourbillon Capital, put it this way earlier this yr at the Milken Institute global conference in Los Angeles: “What do you know that a machine cannot work out?”

FTs at work, not juz beating up locals & stealing their lunches

In Uncategorized on 13/06/2018 at 4:10 am

What TOC, TRE and other anti-PAP sites, and even mothership don’t tell S’poreans about the great things FTs are doing here:

In a recent ranking of the most cited artificial intelligence research papers, which was studded with the likes of MIT and Google, a perhaps surprising name stood out: Nanyang Technological University. In fact, the Singapore university ranked second in the top ten only to Microsoft.

https://www.ft.com/content/4fb6269c-696b-11e8-8cf3-0c230fa67aec

Morocco Mole, Secret Squirrel’s sidekick, tells me that I highlighted what makes NTU great in AI in NTU’s global first in AI. And that these researchers are all FTs, though Secret Squirrel says there’s a true blue S’porean among the lot, but he’s not sure.

More of these FTs please.

 

NTU’s global first in AI

In Uncategorized on 28/04/2018 at 11:04 am

Did u know that artificial-intelligence researchers in Singapore have managed to teach industrial robots to assemble an IKEA chair? And that this is no mean feat? It’s a global first.

More details from the Economist

In a paper just published in Science Robotics, a group of researchers at Nanyang Technological University, in Singapore, report having managed to get a pair of ordinary industrial robots to assemble most of a piece of flat-pack IKEA furniture.

The chair in question was a model called STEFAN. The robots’ job was to assemble the frame. This requires several pieces of dowelling to be inserted into pre-drilled holes before the parts are pressed together. In total, says Pham Quang Cuong, one of the paper’s authors, 19 components are involved.

The robots were off-the-shelf arm-shaped machines of the sort found in factories around the world, combined with a stereoscopic camera that can produce three-dimensional images. A pair of videos released by the researchers show the robot arms making various mistakes, dropping dowelling on the floor or misaligning components, before succeeding at their task after almost nine minutes of slow, careful work.

Even with that abundance of caution, though, the robots needed quite a bit of hand-holding. They were given precise instructions before they started (along the lines of, “Arm 1: take the side piece. Arm 2: grab a dowel. Arm 1: rotate side piece so that hole is pointing up. Arm 2: insert dowel into top-left hole.” And so on.). Before the nine minutes of assembly began, the robots spent a further 11 minutes scanning their environments and planning the movements needed to carry out these instructions, before they tried to execute them. Moreover, though the larger components of the chair were scattered around at random, meaning the robots had to use the camera to identify them by comparing them with electronic representations loaded into a database, the dowels were gathered together and placed upright in a container.

The result is, nevertheless, sufficiently impressive, says Dr Pham, for his research group to have received considerable interest from industry. In future he and his colleagues hope, gradually, to remove the robots’ training wheels. One idea is to get the machines to learn what to do for themselves by watching a human being assemble the chair. Given the difficulties that many people apparently have with IKEA’s products, that may, however, also teach them how to toss the whole thing aside in frustration.

https://www.economist.com/news/science-and-technology/21740733-cower-your-silicon-overlords-puny-humans-robots-can-assemble-ikea

But a human could do it a lot faster, leading the Economist to say this

highlights a deep truth about the limitations of automation. Machines excel at the sorts of abstract, cognitive tasks that, to people, signify intelligence—complex board games, say, or differential calculus. But they struggle with physical jobs, such as navigating a cluttered room, which are so simple that they hardly seem to count as intelligence at all. The IKEAbots are a case in point. It took a pair of them, pre-programmed by humans, more than 20 minutes to assemble a chair that a person could knock together in a fraction of the time (see article).