Connect with us

Artificial Intelligence

Panelist at Information Technology and Innovation Foundation Event Say Deepfakes Are a Double-Edged Sword



Screenshot of Panelists at IFIT Deepfakes Discussion

WASHINGTON, March 23, 2020 – “Democracies will be at a disadvantage” if deepfakes become common, Emerging Technologies Fellow Lindsay Gorman told panelists gathered online, on March 12 at the Information Technology Innovation Foundation.

Surprisingly, while deepfakes present an extraordinary, fast-moving challenge, deepfakes can be used for good as well.

Gorman argued that because democracies rely on facts and trust in government figures who can communicate directly with the people, deepfakes can erode this system.

Deepfakes are altered “synthetic media” that can take several forms: fake videos, images, comments, texts, audio, etc.

Gorman said there has been an “explosion of cheapfakes” in the political realm. When there is a “technical manipulation” it would be helpful to clearly label it rather than take it down and enter into the discourse of constitutional rights, said Gorman.

Pornographic videos are the most common deepfakes, but the conversation seems to swirl around misinformation, said ITIF Vice President Daniel Castro.

Several universities are working to “help communities in detection” because these technologies are varied and diverse, said Facebook Cybersecurity Policy Lead Saleela Salahuddin. “The speed of detection is critical,” said Salahuddin.

Salahuddin said users have to be cautious because the “adversary will pivot to find some loophole” in your protections and detection.

The audio aspect creates a great threat because the technology can learn the speaker’s voice and regurgitate scripted lines, said Ben Sheffner, vice president and associate general counsel for Motion Picture Association.

Salahuddin said a financial institution Facebook works with was concerned about what deepfake audio could mean for accessing financial information and accounts. Salahuddin suggested that while deepfakes are spreading wildly, this speed could be a catalyst for “growth and resilience.”

“Detection is crucial,” said Gorman. But regarding misinformation, instead of solely being distrustful of everything users see, there is a need to find factual and trustworthy sources of information, Gorman said.

Not trusting anything is an “authoritarian model,” warned Gorman.

Congressional fever to legislate regarding deepfakes can be “tempered” by recognizing that there are laws that already protect users from a lot of the harms that are inflicted by deepfakes, said Sheffner.

But the first amendment makes it challenging to regulate even “false speech,” Sheffner said.

Michael Clauser, head of data and trust at Access Partnership, doesn’t think deepfakes are all bad—a position he came to over time. While working with a technologist, Clauser learned that the “underlying technology” of deepfakes can be used to solve issues by orchestrating and training algorithms when they are missing data.

For example, deepfakes can train artificial intelligence to recognize cancer in diagnostic tests.

Clauser said “draconian laws” for deepfakes would actually be more damaging for the United States than helpful.

Adrienne Patton was a Reporter for Broadband Breakfast. She studied English rhetoric and writing at Brigham Young University in Provo, Utah. She grew up in a household of journalists in South Florida. Her father, the late Robes Patton, was a sports writer for the Sun-Sentinel who covered the Miami Heat, and is for whom the press lounge in the American Airlines Arena is named.

Artificial Intelligence

CES 2022: Artificial Intelligence Needs to Resonate with People for Widespread Acceptance

Even though stakeholders may want technologies that yield better results, they may be uncomfortable with artificial intelligence.



Pat Baird speaking at CES 2022

LAS VEGAS, January 6, 2022 – To get artificial intelligence into the mainstream, the industry needs to appease not just regulators, but stakeholders as well.

Pat Baird, regulatory head for software standards at electronics maker Philips, said at the Consumer Electronics Show Thursday that for AI technology to be successfully implemented in a field like medicine, everyone touched by it needs to be comfortable with it.

“A lot of people want to know more information, more information, more information before you dare use that [technology] on me one of the members of my family,” Baird said, “I totally get that, but it is interesting – some of the myths that we see in Hollywood compared to how the technology [actually functions],” adding to be successful you have to win the approval of all stakeholders, not just regulators.

“It is a fine line to take and walk,” Baird said. “I think we need to make sure that the lawmakers really understand the benefits and the risks about this – not all AI is the same. Not all applications are the same.”

Like accidents involving autonomous vehicles, rare accidents for AI can set the technology back years, Baird said. “One of the things that I worry about is when something bad happens that’s kind of reflected on the entire industry.”

Baird noted that many people come prepared with preconceived biases against AI that make them susceptible to skepticism or hesitancy that a technology is safe or will work.

But he did not go so far as to say these biases against AI are putting a “thumb on the scale” against AI, “but [that thumb] is floating near the scale right now.”

“That is one of the things that I’m worried about,” he said. “Because this technology can make a difference. I want to help my patients, damn it, and if this can only improve performance by a couple percent, that is important to that family that you just helped with that [technology].”

Joseph Murphy, vice president of marketing at AI company Sensory Inc., said, “Just like everything in life it’s a tricky balance of innovation, and then putting up the speed bumps to innovation. It’s a process that has to happen.”

On Wednesday, Sally Lange Witkowski, founder of business consulting firm Slang Consulting, said that companies should be educating consumers about the benefits of 5G for widespread adoption.

Continue Reading

Artificial Intelligence

Henry Kissinger: AI Will Prompt Consideration of What it Means to Be Human

Event with the former Secretary of State discusses our current lack of knowledge on how to responsibly harness AI’s power.



Former Secretary of State Henry Kissinger

WASHINGTON, December 24, 2021 – Former Secretary of State Henry Kissinger says that further use of artificial intelligence will call into question what it means to be human, and that the technology cannot solve all those problems humans fail to address on their own.

Kissinger spoke at a Council on Foreign Relations event highlighting his new book “The Age of AI: And Our Human Future” on Monday along with co-author and former Google CEO Eric Schmidt in a conversation moderated by PBS NewsHour anchor Judy Woodruff.

Schmidt remarked throughout the event on unanswered questions about AI despite common use of the technology.

He emphasized that the computer systems may be able to solve complex problems, such as in physics dealing with dark matter or dark energy, but that the humans who built the technology may not be able to determine how exactly the computer solved the problems.

Along the lines of this potential for dangerous use of the technology, he stated how AI development, though sometimes a force for good, “plays” with human lives.

He pointed out that to deal with this great technological power, almost every country now has created a governmental to oversee the ethics of AI development.

Schmidt stated that western values must be the dominant values in AI platforms that influence everyday life such as ones that have key implications for democracy.

With all the consideration on how to make AI work so it is effective but also utilitarian, Kissinger noted how much human thinking must go into managing the “thinking” these machines do, and that “a mere technological edge is not in itself decisive” in terms of AI that can compete with adversaries such as China’s diplomatic technological might.

Continue Reading

Artificial Intelligence

Vaccine Makers Promote Use of Artificial Intelligence for Development

Artificial Intelligence assists in the development of vaccine research and trial testing, makers say.



Najat Khan, Janssen’s research and development global head of strategy

WASHINGTON, December 15, 2021 – Artificial intelligence is helping accelerate the development of COVID-19 vaccines.

Leaders in Janssen’s and Moderna’s research and development groups said Tuesday that AI will help drug makers create better, more effective vaccines for patients.

Speaking at Bloomberg’s Technology Summit on Tuesday, Najat Khan, Janssen’s research and development global head of strategy, said AI is speeding up the delivery of new vaccines for populations in need. (Janssen is a subsidiary of Johnson & Johnson.)

“We use AI and machine learning to predict performance of clinical sites for potential [vaccine] trial sites,” Khan said. AI can help researchers target patients for trials to obtain more comprehensive data sets. Vaccine developers spend time, money, and resources finding patients to participate in clinical trials.

Khan said “only four percent” of eligible patients join a clinical trial. AI can help researchers focus their efforts to identify patients to participate, she said.

Outstanding concerns with AI

Despite AI’s usefulness in vaccine development, Khan said there is still a gap that exists between the information available in healthcare and what’s useful for AI. “There’s lots of data generated in health care, but it’s not connected,” Khan stated. “If it’s not connected, it’s fragmented.”

The problem, Khan said, is the varying systems health clinics use to input and store patients’ information. “Different systems across different clinics needs the same data,” Khan added. “I can go to two different clinics, each one year apart, and my data would be separate.”

On a large scale, mismatched datasets lead to “an over-index of patient information in some areas and an under-index in others,” she said.

For better innovation in treating and curing diseases, health providers need better ways to gather share data while complying with patient privacy concerns, Khan added.

One of health care providers’ challenges is effective data minimization and ensuring that health entities only use patient data according to the patient’s consent over the use of their data. The industry is starting to see progress with tokenization, Khan said, which anonymizes data and links with other data sources for a specific patient-focused purpose.

“This allows us to do even more with AI,” Khan said.

Continue Reading


Signup for Broadband Breakfast

Get twice-weekly Breakfast Media news alerts.
* = required field