Connect with us

Tech

How AI iteration can uplevel the customer experience

Avatar

Published

on

Watch the Low-Code/No-Code Summit on-demand sessions to learn how to successfully innovate and achieve efficiencies by educating and scaling citizen developers. look now.


We love stories of dramatic breakthroughs and tidy endings: the lonely inventor masters the technical challenge, saves the day, the end. These are the recurring tropes surrounding new technology.

Unfortunately, these tropes can be misleading when we are actually in the midst of a technological revolution. It’s the prototypes that get too much attention rather than the complex, incremental refinement that truly delivers a breakthrough solution. Take penicillin. Discovered in 1928, the drug didn’t actually save lives until it was mass-produced 15 years later.

story is so funny. We love our stories and myths about groundbreaking moments, but often the reality is different. What’s really happening—those often long periods of refinement—make for far less exciting stories.

Here we are in the Artificial Intelligence (AI) and Machine Learning (ML) Place. Right now we are seeing the excitement of innovation. There were amazing prototypes and demos of new AI language models like GPT-3 and DALL-E 2.

incident

Intelligent Security Summit

On December 8th, learn about the critical role of AI and ML in cybersecurity and industry-specific case studies. Sign up for your free pass today.

Join Now

Despite the buzz they’ve caused, these types of big language models have yet to revolutionize industries — including ones like customer support where the impact of AI is particularly promising, let alone general business cases.

AI for Customer Experience: Why Aren’t Bots Having More Impact?

News of new prototypes and tech demos often focus on the model’s “best case” performance: What does the golden road look like when everything works perfectly? This is often the first indication that disruptive technologies are coming to market. But contrary to intuition, for many problems, we should be much more interested in the “worst-case” performance. Often the lowest expectations of what a model will achieve are far more important than the higher ones.

Let’s look at this in the context of AI. A Customer Support Bot the sometimes gives customers no answers, however never gives them misleading ones is probably better than a bot that always answers but sometimes gets it wrong. This is crucial in many business contexts.

That doesn’t mean the potential is limited. An ideal state for AI customer support bots would be to be able to answer many customer questions—ones that don’t require human intervention or nuanced understanding—”free form” and correctly 100% of the time. This is rare today, but there are disruptive applications, techniques, and embeds working toward it, even in today’s generation of support bots.

But to get there, we need easy-to-use tools to get a bot up and running, even for less-technical implementers. Luckily the market has matured in the last 3 to 5 years to get us to this point. We are no longer dealing with an immature bot landscape where there are only Google DialogFlow, IBM Watson and Amazon Lex – good NLP bots but very difficult to use for non-developers. It is the ease of use that turns AI and ML into an adaptable and impactful product.

The future of bots isn’t a new, flashy use case for AI

One of the biggest things I’ve learned from seeing companies deploy bots is that most don’t get the deployments right. Most companies build a bot and let it try to respond customer questions, and watch it fail. That’s because there’s often a big difference between an account manager getting their job done, and articulating it correctly enough that something else — an automated system — can do it, too. We typically see that businesses need to iterate to achieve the accuracy and quality of bot experience they initially expect.

For this reason, it is critical that companies do not rely on scarce developer resources as part of their iteration loop. Such trust often results in an inability to achieve the actual standard that the company wanted, leaving a poor-quality bot that hurts credibility.

This is the main component of this complex, incremental refinement that doesn’t produce exciting stories, but delivers a real, game-changing solution: bots must be easy to build, iterate on, and implement – independently, even by people who aren’t trained in engineering or development .

This is not only important for usability. There is another consideration at play. When it comes to bots that answer customer support questions, our internal research shows we’re facing one Pareto 80/20 Dynamic: Good information bots are already about 80% where they will ever go. Instead of trying to squeeze out the last 10-15% of information requests, the industry’s focus must now shift to figuring out how to apply the same technology to solve the non-information requests.

Democratize trading with no-code/low-code tools

For example, in some business cases it is not enough just to provide information; a action (e.g. reschedule an appointment, cancel a booking, or update an address or credit card number). Our internal research has shown that the percentage of support calls that require action reaches a median of around 30% for organizations.

It needs to be easier for companies to actually set up their bots to perform these actions. It has something to do with that No code/low code Movement: Because developers are scarce and expensive, actually enabling the teams most responsible for bot implementation to be able to iterate without dependencies is disproportionately valuable. This is the next big step for business bots.

AI in Customer Experience: From Prototypes to Opportunities

A lot of attention is paid to prototyping new and emerging technologies, and right now there are new and exciting developments that will make technologies like AI, bots and ML and the customer experience even better. However, the clear and present opportunity for organizations is to continue to improve upon and iterate on the technology already established – to leverage new product capabilities to integrate this technology into their operations so they can realize the business impacts already available.

We should focus 80% of our attention on delivering what we already have and only 20% of our time on the prototypes.

Fergal Reid is the Head of Machine Learning at Intercom.

data decision maker

Welcome to the VentureBeat community!

DataDecisionMakers is the place where experts, including technical staff, working with data can share data-related insights and innovations.

If you want to read about innovative ideas and up-to-date information, best practices and the future of data and data technology, visit us at DataDecisionMakers.

You might even consider contribution of an article of yourself!

Read more from DataDecisionMakers

Tech

ICA Miami and other museums continue to collect and display NFTs even as the market sinks, arguing digital art has a long history and valuations are unimportant (Farah Nayeri/New York Times)

Avatar

Published

on


Farah Nayeri / New York Times:

The ICA Miami and other museums continue to collect and display NFTs even as the market falls, arguing that digital art has a long history and reviews are unimportant– ICA Miami owns some, and MFA Boston and others are selling them to reach new audiences “in a world where Instagram and Siri are rampant.”

Continue Reading

Tech

Elon Musk Hopes to Test a Brain Implant in Humans Next Year

Avatar

Published

on

In a presentation showcasing the Neuralink implant that Elon Musk hopes will one day connect the human brain to a computer, two monkeys reportedly moved computer cursors with their brains.

The trick was first documented by others in a human in 2006, in the pre-YouTube era, and with much more cumbersome technology that tethers patients to a computer with a cable.

Mr. Musk’s presentation Wednesday night offered little that was significantly new from previous demonstrations of the device. He further claimed that the implant could allow people with paralysis to have computer control outside of a laboratory setting. However, experts in the field questioned whether the demonstration showed much progress with the device, particularly given the breadth of work underway across the country.

“These are incremental advances,” said Daniel Yoshor, a neurosurgeon and neuroscientist at the Perelman School of Medicine at the University of Pennsylvania, who has worked with similar devices after seeing the presentation. “The hardware is impressive but does not represent a dramatic advance in restoring or improving brain function.”

Neuralink is not licensed by the Food and Drug Administration to sell the device. Mr Musk said on Wednesday that the company submitted most of its paperwork to the agency in order to get permission to implant its device in a human. He predicted human testing in six months, but any move toward human trials would be up to the FDA after a full assessment of the risks of surgical implantation and the safety of the device.

Neuralink originally scheduled the event for late October before Mr. Musk, a multi-billionaire, postponed the presentation amid one of the more chaotic months of his career. He recently completed his on-and-off purchase of Twitter, which has garnered much of his attention — and sparked significant controversy — over the social media company’s management.

As Mr. Musk juggles these and other responsibilities — he also oversees electric car maker Tesla and rocket company SpaceX — Neuralink emerges from a time of transition. Last year, Max Hodak, the company’s president and one of its co-founders, left the company to start his own business in this space. Neuralink’s CEO is officially Jared Birchall, a money manager who runs Mr. Musk’s family office.

Wednesday night’s presentation focused on the “Link” device, which resembles an inch-wide stack of multiple coins with hundreds of hair-thin threads. According to Mr. Musk’s 2020, a surgical robot would cut a hole in the skull and push the electrode threads into the brain’s gray matter company presentation. The coin-like piece would sit flush with the skull.

Leading companies in the field of brain-computer interface technology have closely watched Neuralink’s investment in a device that works without protruding cables or hardware. But Mr Musk’s presentations so far have unsettled and overwhelmed many of them.

A 2021 Neuralink presentation of a monkey playing the video game Pong with its mind was similar to a primate demonstration at Brown University in 2001, although it had a far clunkier system.

In a 2020 presentation Showing off a pig with the implant, Mr Musk suggested the device could “tackle” conditions like paralysis and insomnia, and even give a user “superhuman vision.” Such applications sound like science fiction to scientists focused solely on restoring basic functions like typing, speaking, or lifting a fork to those who lost them after a spinal cord injury or a nasty diagnosis. For such patients, the benefits outweigh the small but serious risk of brain surgery.

“No one talks about implanting healthy people,” said Cindy Chestek, associate professor of biomedical engineering at the University of Michigan, whose lab works on restoring function to amputees.

On Wednesday night, Mr. Musk said plans for his device included giving the blind sight and giving “full-body functionality” to someone with a severed spinal cord. The claims were applauded by the audience but do not reflect the current state of affairs.

“I wouldn’t say that with confidence,” said Dr. Yoshor after Mr. Musk claimed the Neuralink device would see people who had never seen before. “For a patient with congenital blindness, I would be very insecure about this type of device.”

Safety will be the FDA’s primary concern when considering whether the device could be tested in humans, said Cristin Welle, an associate professor of neuroscience at the University of Colorado who helped draft FDA guidance on brain-computer implants before she left the agency in 2016.

dr Welle said regulators will focus on whether the device would damage the brain or pose undue risks to patients. She said the device’s durability is also being considered, as brain fluids may eat through the insulation that coats the hundreds of hair-like electrodes on the Link device.

So far, Neuralink has tested the device on sheep, pigs and primates, according to records filed with the Department of Agriculture.

Several other companies and scientists have already received FDA approval to study similar devices in humans. In 2004, researchers conducted human trials with the Utah array, a device the size of a baby aspirin pill that is spiked and surgically placed on the brain. It is connected by a cable to a small computer installed on the head, which sends data to a computer. This neural interface system is called BrainGate.

With the pieces in place, scientists are looking for patterns in neurons’ electrical current that signal the brain’s intent to type letters or raise a hand. The code, in turn, commands a computer or robot to perform the task.

Nearly three dozen patients were tested with the Utah array device. Using technology, people with paralysis or other disabilities lifted a cinnamon coffee with a robotic arm in 2011, typed letters quoting Shakespeare, and lifted in 2012 forks Mashed potatoes in 2016.

But the Utah array is not designed for continuous use. Rising from the skull, it ties users to a cable connected to a computer and puts them at risk of brain infection. For these reasons and others, companies like Neuralink are working to build devices that are fully implanted.

Continue Reading

Tech

Elon Musk claims Neuralink is about ‘six months’ away from first human trial

Avatar

Published

on

Elon Musk said at a Show and Tell event on Wednesday that his brain-computer interface company, Neuralink, could implant one of its devices in someone’s head within the next six months — which means it’s not happening this year. He also claimed that he would have the device implanted in his own head at some point in the future.

During the presentation, Musk said the company has submitted most of the documentation needed for a human clinical trial to the Food and Drug Administration, which regulates medical devices in the United States. Musk previously said he was hoping for human trials to begin in 2020 and then 2022. Now that has been postponed to at least 2023.

Neuralink’s goal is to develop a device that can be implanted in the brain and use it to control a computer with brain activity. Musk announced back in 2019 that the company was tests his device on monkeys. 2020 is it trotted pigs with the implants. And last year, Neuralink released a video shows a monkey playing pong with his brain. This year the monkeys are back. In a video demonstration, one of them helped “type” the phrase “Welcome to show and tell” with her implant by focusing on highlighted words and letters. Another video showed the monkeys being trained to charge the devices by sitting under a wireless charger.

A monkey sucks on a banana smoothie straw placed so its head bumps against an embedded wireless charger in a branch.

The wireless banana smoothie charger.
Screenshot by Sean Hollister / The Verge

Later in the presentation, the Neuralink researchers also showed a pig on a treadmill, which they said will help them study how to address mobility issues in humans in the future.

The Neuralink devices themselves are small, with multiple flexible “threads” that can be inserted into the brain. “It’s like replacing a piece of your skull with a smartwatch, for lack of a better analogy,” Musk said.

In about 15 minutes, 64 of these “threads” can be implanted in the brain using a robotic system, DJ Seo, the vice president of Implant and co-founder of Neuralink, said during the presentation — while using a mannequin to show how the process works could work.

Threads by Neuralink.

Threads by Neuralink.
Screenshot by Sean Hollister / The Verge

The reason for the robotic surgeons is because of how tiny these threads are. “Imagine taking a hair from your head and sticking it in jelly covered with cling film, doing it with a precise depth and precision, and doing it 64 times in a reasonable amount of time,” said Christine Odabashian, the director by Neuralinks hardware insertion team.

The company’s 2019 and 2020 demos were designed as recruitment events, and this one is no different; The company admitted that recruitment was the main goal of the evening. Neuralink is currently trying to fill many different types of jobs as it moves from “prototype to product,” Musk said on today’s Show and Tell.

The event consisted mainly of a technical presentation of the device, showing how the system was built, what challenges the team faced, how the technology has improved so far and what developments are coming next. Company researchers said they are developing treatments that could either help improve or restore vision, or restore movement in people with paralysis. On the tech side, the company has ambitions to ensure that the device itself can be easily upgraded.

“I’m pretty sure if an iPhone 14 was available, you wouldn’t want an iPhone 1 in your head,” Musk said.

Continue Reading

featured