|
|
Patent Chief Questions AI Inventiveness; Renault Takes a Spin; Tech Compensation Soars
|
|
|
|
|
|
|
Andrei Iancu, director of the U.S. Patent and Trademark Office. PHOTO: JAY PREMACK/U.S. PATENT AND TRADEMARK OFFICE
|
|
|
Reporter's Notebook: U.S. patent chief doubtful about AI as inventor. Patent offices around the world have been asked to recognize artificial-intelligence systems that develop new products as inventors, we reported last week. But at least one patent official doesn’t believe AI is capable of inventing, Jared Council reports.
"I don't yet fully subscribe to the view that the machine is completely autonomous and operates without human intervention,” Andrei Iancu, director of the U.S. Patent and Trademark Office, said in remarks that haven't been reported before. “At least as of today, and probably the foreseeable future, the AI machine is just another tool.”
[continues below]
|
|
|
|
A Missouri-based AI expert is arguing the exact opposite in patent applications filed in the U.S., Europe and Israel listing his AI system as the sole inventor of two new products despite patent laws that state only humans can be listed. Lawyers involved in the applications said the AI architect didn't come up with the ideas for the two products—a specially designed container lid and a flashlight for drawing attention in emergencies—and didn’t direct the machine to invent them.
In the U.S., legally recognizing AI as an inventor will ultimately be up to federal courts or Congress, but Mr. Iancu is helping formulate policy positions for his office that could influence those opinions. The U.S. patent director declined to comment on specific cases, but said his office is gathering public feedback on the topic.
Mr. Iancu said the court system has established that the critical aspect of being an inventor is the act of conception, and that only human beings—as opposed to corporations—can conceive an idea. When it comes to AI systems, humans have to decide on the training data and set parameters on what the machine can do, he said, which essentially makes AI a tool.
The advent of photography raised similar questions about who gets credit for the output of a machine, he said.
"Perhaps that's the kind of situation we are in now, although with a more advanced tool,” Mr. Iancu said. “Or, perhaps now we are in a completely new world, where a new consideration needs to be given because the man and the machine in this example have separated more than in the past.”
|
|
|
Probing the ethics of AI in health care. The rise of artificial intelligence poses a range of ethical uncertainties and dilemmas for scientific researchers, business people and the government. The Wall Street Journal recently invited three experts to participate in an email exchange addressing those ethical problems.
Here are responses to one of the questions up for discussion: Could AI undermine physicians’ discretion?
|
|
One interesting question has to do with how, in the future, AI will interact with health-insurance reimbursement. Physicians may retain the discretion to deviate from AI recommendations. But that may not matter if the insurer is allowed to say we will only reimburse for what the AI recommends. — Glenn Cohen, a professor at Harvard Law School and director of its Petrie-Flom Center for Health Law Policy, Biotechnology and Bioethics
|
|
|
AI systems might take on an authority they don’t deserve. There is a likely scenario where AI systems move into a role of controlling clinicians’ decisions and workflow, as the electronic medical records prompts and warnings are already starting to do. This is great when it can be used to demonstrably reduce errors and improve health outcomes; not so great if the chief utility is improving profit or finessing evaluation metrics. Deviating or challenging algorithmic recommendation is going to be challenging. — Danton Char, assistant professor of anesthesiology, perioperative and pain medicine at Stanford University Medical Center
|
|
|
Everyone agrees that algorithms are going to be a big part of medicine in the future, but we aren’t developing the human-capital pipeline against that goal. We don’t see it reflected in our pre-med requirements, or medical-school curricula. — Ziad Obermeyer, acting associate professor of health policy and management at the University of California, Berkeley
|
|
|
|
|
|
|
A Renault EZ-GO autonomous electric concept car presented at the 2018 Paris Motor Show. PHOTO: AGENCE FRANCE-PRESSE/GETTY IMAGES
|
|
|
Renault has started a public trial of an autonomous, electric and shared on-demand car service, CNBC reports.
The trial started Monday and will last until Nov. 8, with roughly 100 people using the service, according to CNBC. Highlights from the story:
-
The test will take place on the Paris-Saclay “urban campus” in France and will use the prototype Renault ZOE Cab. Two vehicles with differing features will be deployed, with passengers able to hail them using a mobile app. The campus includes academic research centers of the Université Paris-Saclay, as well as startups and research and development centers of multinational firms.
-
The announcement follows last week’s news that Renault, together with Waymo and authorities from the Paris region, would explore the creation of an “autonomous mobility service” between Roissy-Charles de Gaulle Airport and the business district of La Défense. Renault is one of many companies testing autonomous vehicles.
|
|
|
Megvii Technology pushes ahead with IPO. The Beijing-based facial-recognition startup is still planning an initial public offering in Hong Kong and wants to hold a listing hearing next month, despite being blacklisted by the Trump administration, Bloomberg reports, citing unnamed sources. Megvii has lost access to U.S. technology including Nvidia Corp. chips that are key to its business, and is appealing the U.S.'s decision to place it on the Commerce Department's Entity List, Bloomberg writes.
|
|
|
|
Compensation soaring for developers, engineers in AI. Six-figure bonuses, outsize equity stakes and the flexibility to work from just about anywhere are some of the perks companies are offering IT workers as they compete for talent in a tight labor market, reports WSJ’s Angus Loten.
Gartner Inc. estimates that most large U.S. companies are competing to fill many of the same technology roles, including computer and information research scientists, systems managers, analysts, engineers and software architects.
Demand for these workers is growing as companies world-wide seek an edge over competitors by using technology such as cloud computing, data analytics and artificial intelligence. Sought-after IT workers, such as developers and engineers with experience in AI, who take new jobs this year are fetching average annual salaries of $200,000, up from $150,000 a year ago, according to staffing firm Mondo, a subsidiary of Chicago-based Addison Group LLC.
|
|
|
|
|