AI's new painter, Ai-Da can paint human faces
The robot will be able to sketch human faces. (Image Credit: Reuters)
There's a chance some people have always wondered how they may appear to the eyes of a robot. Now they may be able to see that. Engineers are working closely with an art gallery owner to create a robotic artist. Ai-Da, the artificial intelligence robot will have a bionic hand that sketches what it sees. The bot will use AI technology to carry out its tasks - with cameras inserted in the eyeballs to see what's in front of it and AI-powered computer vision to identify humans, copy their expressions and track their features. The bot has a realistic human face with silicone skin, a moveable mouth, eyes, teeth and gums and a voice to answer any question. Ai-Da won't have the same artistic skills as a skilled human artist, but the creation will put a human-like face in the field of AI painting.
New initiative for AI set by White House
White House image. Trump's latest initiative brings more development for AI. (Image Credit: Joyve N. Boghosian)
After dismissing AI subject talks during Trump's SOTU address, a new directive could help shape everything from healthcare to military warfare. Trump's new AI initiative will have a showing in federal agencies to prioritize AI in their research and development, but there is no specific funding for these areas. The initiative also includes efforts to help educate American workers by using a number of programs led by the National Council for the American Worker and AI fellowships care of federal agencies. It also aims to gather more support from the government to give access to cloud computing and data-sharing for new AI systems. The order comes from concerns from China wanting to be a global AI superpower. The initiative also asks federal branches o establish a regulatory framework for AI in industrial sectors.
AI and strain engineering focuses on electronic properties
Strain on crystalline properties can help to create new semi-conductive technology for the future. (Image Credit: Dennis M. Rivera Pichardo)
Applying a bit of strain to a crystalline material's properties can alter it dramatically but finding the right strain for it can be a challenging task. Which is where AI can come into play with this. An international team of researchers has come up with a way for machine-learning to produce the best results. The neural network algorithm estimates how the direction and degree of strain can affect the main property conducting the efficiency of semiconductors. This makes them more efficient without educated guesses from humans. This could lead to an increase in inventions involving semiconductors that are more powerful, but with small changes. A silicon solar cell would be able to harness as much energy but with a smaller size at a thousandth the thickness. Diamonds could also have a role in this by replacing silicon in processors, making speeds up to 100,000 times faster. AI can also be used for thermal and optical traits but that is more challenging because of the complexity level needed for chips.
AICAN is AI's latest in identifying paintings from 500 years ago
The paintings can be recognized by the AI system, AICAN which then tells us what the painting is. (Image Credit: Rutgers University)
Artificial Intelligence has taken shape into the art scene over the past few years - with tech giants and everyday artists playing around with the technology. AICAN (Artificial Intelligence Creative Adversarial Network) is an autonomous system designed to recognize over 500 years' worth of Western art, and it comes up with its own interpretations of these pieces. Adversarial Network is very different from GANs (Generative Adversarial Networks) and works from two sets of nodes. One set generates images based on the visual training data set while the other node judges how closely the generated image matches the image from the training data. AICAN isn't trained on a certain set of visuals either, but instead, uses aesthetics of Western art history as it goes through databases, taking in examples of everything it sees without any preference to a subject or genre. To date, the system has found over 100,000 examples. This system isn't meant to replace human artists, either.
AI vs. human in debate - human wins
Watch Miss Debater lose a debate to its human opponent.
AI has also gained the ability to participate in debates, but the robot lost the battle. Champion debater Harish Natarajan came out as the victor against IBM's Miss Debater AI at the conference in San Francisco. The debate went on for 25 minutes on pre-school subsidies. Both sides were given 15 minutes to prepare - they then had a 4-minute opening statement, 4-minute rebuttal and a 2-minute closing summary. The AI system, Miss Debated gathered info from its database by gathering 10 billion sentences from newspapers and academic journals. A female voice sourced from the human-sized black box gave out its answers while three blue balls floated around. Debating may prove to be more challenging in the future because it's a lot more complicated than Go and video games. For a debater to be successful, they must have three parts - they need to construct a lot of information and come up with arguments, it involves explaining things in a clearly structured way and they need to make those arguments matter to the audience. All of which need the use of language, emotion, rhetoric and examples. A machine may know the first one, but the other two will be challenging.
Have a story tip? Message me at: cabe(at)element14(dot)com
http://twitter.com/Cabe_Atwell