[ad_1]
Comment In the future, AI may be a challenge for US Patent and Trademark Office (USPTO) officials, who have to wrap their heads around complex technology that does not fully comply with today’s laws.
Under the Department of Commerce, the main mission of the USPTO is to protect intellectual property or IP. Creators aim to prevent patent applications from copying their competitors’ licenses, and allow creative businesses to enrich themselves with new designs rather than hindering large-scale innovations.
Fast-paced technologies such as in-depth learning are pushing the limits of today’s IP policies and regulations. Authors are trying to apply traditional patent control rules to simple machine-learning innovations, and bad decisions can hinder competition between public and private AI creators. We all know how much ownership of software and other technologies goes beyond USPTO and will cause headaches in the years to come.
“AI is already impacting many industries and many aspects of our society,” said Katie Vidal, director of the agency and former engineer at the opening of the AI and Emerging Technologies (ET) partnership almost last month.
“AI and new technologies have the potential to dramatically improve our daily lives. Here, not only in the United States, but also around the world, they offer innumerable and unpredictable benefits to our social security. But the bottom line is, we have to get it right.
“We must ensure that we are developing laws, policies and practices that benefit the United States and the world.”
Copyright disseminates valuable knowledge, and offers ideas on how engineers and scientists can advance technologies or innovate. Creators must meet a list of criteria for their applications to be viewed. They must not only show that their creations are new, vague, and useful, but that they can express their work in a way that can be understood and replicated by someone skilled in the same field.
And here is the massage.
Nerve networks cannot be easily defined. The process of numerical manipulation, which seems to translate input information into magic, is often vague and cannot be interpreted. Experts often do not know why the model works this way, making it difficult for proprietors to review the app’s Nitty-Gritty details.
Reproduction is also very difficult to learn by machine. Developers need access to redesign a model’s training data, metrics, and / or weights. Submitting this information with a patent application may satisfy the challenger, but it may not be for the benefit of the creators or the general public.
Medical information taken from real patients, for example, can open up all kinds of risks if it is given to train, publish and store government agency staff an algorithm that can detect tumors. Public disclosure of the system may reveal ownership information. In some cases, it may be easier not to own the technology.
The USPTO has previously faced obstacles in enforcing patents on AI innovations. Mary Critaris, USPTO’s chief policy officer and director of international affairs, noted that the US patent was revoked in 2014 following a US Supreme Court ruling. [PDF] Alice Corporation vs CLS Bank International Case. Judges ruled that CLS could not infringe on Alice’s financial computer software, because it was too subtle.
Like the laws of nature and natural phenomena, abstract ideas cannot normally be patented. The Supreme Court’s decision will therefore have a chilling effect on AI patent applications and admissions, as they are also considered to be very abstract until at least further guidance is given to patent examiners on how to handle draft designs.
“[The data] “There is some evidence that Alice’s decision has had an impact on AI technologies,” Critaris said.
“Allowance rates remain below non-AI application rates until 2019. For this reason, the 2019 USPTO has issued an updated subject competency guideline,” she continued, citing the advice outlined here. [PDF].
“I think that’s why we see an increase in allowances, but Alice’s decision certainly had an impact on AI-related applications.”
As machine learning improves and more patents are applied and selected in court, we can see another flood of allowances.
Last year, a group of U.S. senators called on the USPTO to clarify what patent rights are and why they are “inconsistent and transparent in patent eligibility laws.” “Lack of transparency has not only forced investment in important new technologies, but the courts have completely banned protection for certain important innovations in the research, biofarmutical and life sciences industries,” the letter said.
The USPTO’s clear guidelines help encourage creators to successfully apply for patents. But advice only goes so far. The U.S. courts have a final say in these matters.
And, individually, it is not clear whether or not the technologies created by AI are copyrighted. Who owns art, music, or text IP rights created using generic models? These innovations spoil the content and can mimic certain patterns. Do you infringe copyright?
If these models create content, can they be listed as creators? Current US laws, at least, only know the IP, which is made by “natural persons” to the detriment of one. Stephen Teller has filed a lawsuit against Andre Iankun Dabus, a former director of the Copyright Office, for rejecting an application to create a network.
There was no significant commercial application of these technologies, which was a patent war on the sewing machine in the form of what would happen next patent war.
Some legal experts believe that it would be nice if people could start registering patents for automated learning-learning algorithms. These inventions may not be entirely new but they were the way they were created; Will these be accepted or will they be rejected?
USPTO cannot certainly answer all these questions; Some of these cases must be tried and tested in court.
“There haven’t been a lot of litigation on AI yet,” said Adam Mossoff, a law professor at the Anthony Scalia Law School at George Mason University.
“There was no significant commercial application of these technologies. The next patent war was a patent war on sewing machines, and a patent war on fiber optics, and a patent war on disposable diapers and other things. And when that happens, I think we see a real threat here.
UPTSO has been asked to comment on current policies on what innovations can or cannot be patented.
Some people think that the agency is effective in patenting and protecting patents from patrol trolleys, while others disagree that the agency’s framework hinders innovation for small businesses and startups.
Recent Report [PDF] Conclusion from the Agency Everyone agrees on one thing: “The criteria for identifying a patent must be clear, predictable, and consistent.” ®
[ad_2]
Source link
