Tag: ai

Baidu and Nvidia steal your voice

lost voiceChinese internet giant Baidu and its chum Nvidia claim to have developed an artificial intelligence system which it claims, can successfully imitate your voice after hearing you speak for only thirty minutes.

At Baidu World, the company’s annual tech expo, CEO Robin Li launched the AI project entitled ‘Baidu Brain’, a tripartite initiative which it has undertaken in partnership with Nvidia. Baidu Brain is concerned with AI algorithms, computing power, and big data.

Li said that the system had extensive speech synthesis capabilities and could imitate you completely:

“Anyone just records 50 sentences as required in 30 minutes, and our speech synthesis technology could simulate the person’s voice. We could let everyone have their own voice model.”

We have heard this before. AT&T Labs once promised to bring dead celebrities back to life with a “custom voice” product called Natural Voices. The technology was acquired by speech synthesis specialists Nuance who abandoned it.

However it could have more spooky uses. You could, for example, mimic a general giving orders on the phone by building a database of his public speeches.

So far Baidu has not said what it can do in terms of voice cloning but it might be that in a few years you may been to share some passwords so that you know who you are talking to on the phone.


Google using AI to spot cancer

google-apple-maps-eric-schmidtGoogle has teamed up with some British researchers to come up with a way of using AI to  automatically differentiate between cancerous and healthy tissues on patient scans.

The partnership brings together leading clinicians and researchers at University College London Hospitals NHS Foundation Trust (UCLH) with some of the UK’s top technologists at DeepMind Health, which specialises in using machine learning to solve some of the world’s most difficult problems.

At present, it can take clinicians up to four hours to identify and differentiate between cancerous and healthy tissues on CT and MRI scans of head and neck cancer patients. This process, known as segmentation, is particularly difficult in head and neck cancer patients because their tumours are situated in extremely close proximity to healthy structures such as the eyes and nerves.

Before treatment can begin, clinicians identify the cancerous areas on the scans, and the areas that must be protected from radiation. It is essential that cancerous and healthy tissues are identified accurately so that radiotherapy treatment can be effectively targeted, giving the highest radiation dose possible to the tumour, while preserving healthy, surrounding structures and reducing possible side effects.

The purpose of the research collaboration between UCLH and DeepMind is to develop artificial intelligence technology to assist clinicians in the segmentation process so that it can be done more rapidly but just as accurately. Clinicians will remain responsible for deciding radiotherapy treatment plans but it is hoped that the segmentation process could be reduced from up to four hours to around an hour.

The research involves anonymised radiotherapy images of up to 700 former head and neck cancer patients who have consented to their data being used for research purposes.

Dr Yen-Ching Chang, clinical lead for radiotherapy at UCLH, said: “This is very exciting research which could revolutionise the way in which we plan radiotherapy treatment.

“Developing machine learning which can automatically differentiate between cancerous and healthy tissue on radiotherapy scans will assist clinicians in planning radiotherapy treatment. This has the potential to free up clinicians to spend even more time on patient care, education and research, all of which would be to the benefit of our patients and the populations we serve.

“This collaboration also means our patients continue to benefit from the most cutting-edge developments in healthcare technology.”

Professor Kathy Pritchard-Jones, chief medical officer of London Cancer, the integrated cancer system that serves a population of more than 3.5 million, said: “Head and neck cancer is rare and is one of the most complex tumour sites to treat. Therefore, if we can develop technology to assist in planning radiotherapy treatment for these tumours, we would expect that such a breakthrough would be transferable to other types of cancer. This would not only benefit UCLH patients, but patients across the country.”

DeepMind Co-Founder Mustafa Suleyman said: “This real-world application of artificial intelligence (AI) technology is exactly why we set up DeepMind. We’re incredibly excited to be working with the radiotherapy team at UCLH to explore how AI can help to reduce the time it takes to plan radiotherapy treatment for head and neck cancer patients. We hope this work could lead to real benefits for cancer patients across the country and for the clinicians who treat them.”

AI computers will try to hack each other

cybermen__quot_delete_quot__campaign_by_degaspiv-d33hjoaSeven AI computers will have a crackat hacking each other in Las Vegas early next month.

The seven will take part i nDARPA’s Cyber Grand Challenge finals and try to defend themselves and point out flaws without any human control. The object is to show that machines can beat even the best human hackers.

Mike Walker, programme manager for the CGC siad that it was  proof that eventually the entire security life cycle could be automated.

On average,  flaws in software go unnoticed for around 312 days — which hackers can often exploit. And then once those flaws are noticed by a human, they need to be understood, patched, and then released out to the broader community.

The CGC hopes this problem could be fixed within minutes, or even seconds, automatically.

Seven teams of finalists were given a DARPA-constructed computer. Their task was program it to be able to recognize and understand previously-undisclosed software, find its flaws, and fix it. And once the challenge starts, they won’t be able to jump on a keyboard and do anything more.

“The machines have to comprehend the language of the software, author the logic for that software, write their own network clients, And arrive at the path of the new vulnerabilities entirely on their own.”

While they are scanning their own systems for problems, the machines can also scan the other teams’ systems for issues, but they can’t actually hack them.

Walked likened it to calling your shot in a game of pool, without actually hitting the ball.

Instead, they will send a message of sorts to the DARPA referee, who will then go ahead and see if that exploit is correct, or if what was pointed out could crash the other machine.

The first place team will take home $2 million so it is worth a crack.

AI robot makes another run for it

the-great-escapeA Russian AI robot caused an unusual traffic jam last week after it “escaped” from a research lab, and now it seems that the robotic “one who didn’t get away” had another crack at escaping.

Engineers at the Russian lab reprogrammed the intelligent machine, dubbed Promobot IR77, after last week’s incident, but the robot recently made a second escape attempt.

Last week, the robot made it approximately 160 feet to the street, before it lost power and “partially paralyzed” traffic. Promobot, the company that designed the robot, announced the escapade in a blog post the next day.

Some thought it was a promotional stunt, given the fact that the  company’s blog includes photographs of the robot from multiple angles as it obstructs traffic, and the robot’s escape came a week after Promobot announced plans to present the newest model in the company’s series, Promobot V3, later this year.

Engineers were testing a new positioning system that allows the robot to avoid collisions while moving under its own control. But when a gate was left open, the robot wandered into the street and blocked a lane of traffic for about 40 minutes.

The Promobot was designed to interact with people using speech recognition, providing information in the form of an expressive electronic face, prerecorded audio messages and a large screen on its chest. The company has said the robot could be used as a promoter, administrator, tour guide or concierge.

In light of the robot’s recent escapes, and citing multiple changes to the robot’s artificial intelligence, Promobot co-founder Oleg Kivokurtsev said “I think we might have to dismantle it.”

But in its blog post, the company said it considers the escape a successful test of the machine’s new navigation system, because the robot didn’t harm anyone and wasn’t damaged during the getaway.


Apple shows off its lack of ideas


Screen-Shot-2016-06-13-at-2.08.26-PMIn case you were asleep, Apple held one of its Nuremburg style rallies yesterday which was supposed to show off its latest operating systems and products.

While the Tame Apple Press did its best to promote the rally, er WWDC keynote, there was one thing missing from the event – a noticeable, useful and “game changing product.”   You would not know this fact from the press coverage, but scanning a careful eye over what was released, you cannot help being overwhelmed by the loud-sounding nothing.

The headline launch was iOS 10, which was billed as a new major version of its operating system for iPhones, iPads, and iPod Touches. The “major change” turned out to be a redesigned control centre with interactive notifications and improvements to widgets.  3D Touch can now do more: for example, you can press your screen to clear all notifications at once (hooray).

Next up were the improvements made to Apple’s Siri, which make it easier to accidently call the emergency services without any effort.  Apple promises that you can book rides on services like Uber and Lyft, start and pause workouts, and make payments to send money to friends with Square Cash and other services. Siri will be more intelligent in messaging. For example, if a friend asks for the contact information of another friend, Siri can automatically provide the contact. Siri can also pull up calendar availability in message threads and let you do multilingual typing without switching the keyboard language.

While the changes to Siri do fit into the “super,” “cool” and “amazing” category they are behind what rivals are doing in AI.  Apple, after doing so well with Siri, dropped the ball on this one.

Apple also promised more intelligent photo management, including facial recognition technology to identify people in photos. This is something that is done on Facebook and again is nothing special. Photos can also be sorted by automatically generated topics, like “on the water” or “at the beach.”

There has been improvements to maps including the ability to pan and zoom further ahead to see traffic conditions, quick controls to see route details, and locations of food and gas stations on your route. Maps will tell you how much longer it will take to go to any of these places. Of course if you trust Apple Maps after its launch fiasco you might want to replace this with an App with Google or something else a little more reliable.  Most of that functionality can be found for free on Google anyway.

WWDC was also told that Apple Music now had 15 million paying customers. The figure is rather low considering Apple’s fanbase. Jobs’ Mob is not doing as well as it would like in this “so called” new direction. Once again it depended on numbers from China to make up the difference only to find itself banned by the Chinese authorities.

What did Apple tell WWDC, if it had nothing to really sell? Apple Senior VP Craig Federighi made the somewhat scary prediction that “Children of tomorrow will have no understanding of the English language.”  Maybe it was more of a hope you need a mastery of the English language to spot that the worlds coming out of Apple are spinning a lack of substantial product, or innovative strategy.


Watson gets a job as a lawyer

stupid-lawyer1Biggish Blue’s AI supercomputer Watson has just got a job as a bankrupcy lawyer.

Global law firm Baker & Hostetler has bought itself Ross, the first artificially intelligent attorney built by ROSS Intelligence. Ross will be employed in the law firm’s bankruptcy practice which currently employs more than 50 lawyers.

Ross can understand your questions, and respond with a hypothesis backed by references and citations. It improves on legal research by providing you with only the most highly relevant answers rather than thousands of results you would need to sift through.

It constantly monitors current litigation so that it can notify you about recent court decisions that may affect your case, and it will continue to learn from experience, gaining more knowledge and operating more quickly, the more you interact with it.

Andrew Arruda, ROSS Intelligence co-founder and CEO, other law firms have signed for licences with Ross, and more announcements are expected.

It is nice that lawyers will be the first race of sharks to be wiped out by our robotic overlords.  If we could replace politicans next that would be even better.


Developers fear being replaced by robots

Robby the Robot - Wikimedia CommonsA survey of 550 software developers by Evans Data has found that the biggest thing that keeps them awake at night is being replaced by AI.

Nearly a third said that this was the most worrisome thing in their careers – even more than the second-most identified worry, which was that the platform the developer is working on will become obsolete (23 per cent), or doesn’t catch on (14 per cent).

Concerns about A.I. replacing software developers has academic support. A study by Oxford University, The Future of Employment, warned that the work of software engineers may soon become computerised. Machine learning advances allow design choices that can be optimised by algorithms.

These systems can also detect bugs “with a reliability that humans are unlikely to match,” the study said.

The Oxford researchers, Michael Osborne, of Oxford’s Department of Engineering Science, and Carl Benedikt Frey said that big databases of code also offer the eventual prospect of algorithms that learn how to write programs to satisfy specifications provided by a human.

Janel Garvin, CEO of Evans Data, the thought of obsolescence due to A.I., “was also more threatening than becoming old without a pension, being stifled at work by bad management, or by seeing their skills and tools become irrelevant.”

Concerns about A.I. shouldn’t discourage people from becoming developers, “but it does provide a good case for developers to keep on top of the latest development practices,” he said.

Apple buys more intelligence

mybrainhurtsDesperate to find a way out of its ever shrinking share price and desert of ideas, the fruity cargo cult Apple has decided to buy some intelligence.

Apple has bought Emotient which is an artificial intelligence startup that reads people’s emotions by analysing facial expressions.

The tech giant’s plans for Emotient were not immediately clear and Apple is not talking about it other than to say that the deal is going ahead.

Emotient’s software reads the expressions of individuals and crowds to gain insights that can be used by advertisers to assess viewer reaction or a medical practitioner to better understand signs of pain in patients.

The software could be used for anything Applish but it might be part of Jobs’ Mob’s move into the health market. Equally though it could be used by a device to assess a response to a pushed advertisement.

San Diego-based Emotient had previously raised $8 million from investors including Intel Capital.
There are hints that the move might be to counter something that Tesla Motors Chief Executive Elon Musk did late last year by announced $1 billion in funding for an artificial intelligence non-profit called OpenAI. If that sexy technology goes OpenSource then Apple will have to make more open source projects, and we can’t have that.

Zuckerberg wants an AI butler

what-we-learned-about-facebook-ceo-mark-zucke-L-gl5gYRSocial Notworking supremo Mark Zuckerburg wants to build an artificially intelligent assistant in 2016 to help run his home and assist him at work.

While he is not the first rich person to want a butler, Zuckerburg does have the cash to develop one.

Zuckerberg makes a New Year’s resolution every year and revealed his cunning plan in a Facebook post.

“You can think of it kind of like Jarvis in Iron Man,” Zuckerberg wrote.

Zuckerberg will start the project by exploring existing technology, he wrote. He will then begin teaching the technology to understand his voice so that it will learn to control everything in his home, such as music, lights and temperature.

His plans also include teaching the assistant to let friends into his home by looking at their faces when they ring the doorbell, Zuckerberg wrote. The assistant will visualize data to support Zuckerberg at work, he wrote.

“This should be a fun intellectual challenge to code this for myself,” Zuckerberg wrote. “I’m looking forward to sharing what I learn over the course of the year.”

Previously his New Year’s resolutions have involved reading two books every month and studying Mandarin – although why he needed a year to look at a fruit is anyone’s guess.


AI programme passes University Entrance exam

The Japanese National Institute of Informatics claims that its AI program, developed with university and corporate researchers, achieved an above-average score on a college entrance exam for the first time.

The test covered five subjects including maths, physics and English.  The institute wanted to develop an AI by 2021 that was high enough on Japan’s standardised college entrance exam to be accepted into the University of Tokyo, the nation’s top-ranked university.

The AI received a score of 511 points out of 950, above the national average of 416, and did exceptionally well on math and history-related problems, the institute said.

Getting that score means that the AI has at least an 80 percent chance of being accepted by 441 private universities and 33 national universities.

The software program had been “studying” for the exam since 2011 but had below average scores on similar exams in 2013 and 2014.

It has a long way to go before it can apply to the University of Tokyo, whose list of alumni includes Bank of Japan Gov. Haruhiko Kuroda and former Prime Minister Yukio Hatoyama.

The AI was smart enough to answer some of the questions on the physics test, it failed to do so and scored below average because of its insufficient language processing capability.