Chris Benson says Anthropic recently released a paper and blog post on watermarking to detect AI-generated content.
“It was just a few days ago that Anthropic released their paper, uh, along with a blog post on watermarking, um, so that you can detect”
Practical AI · 0:30:27
Chris Benson says China is currently the powerhouse of open models.
“there's a lot out of China, which is kind of the the the powerhouse planet of open models now.”
Practical AI · 0:09:18
Chris Benson says American-based closed models accessed through APIs still lead their category, while Chinese models are the strongest for self-hosting.
“if you're willing to do APis for closed models, uh, the the the American-based ones, uh, are still leading in that space.”
Practical AI · 0:09:43
Chris Benson describes an agent harness as an operating system for AI that accesses the models and manages and orchestrates the agents.
“I've heard it described as an operating system for AI, where it is it is accessing the models that you're using”
Practical AI · 0:11:22
Chris Benson distinguishes a swarm of agents from a fleet, saying a swarm is usually a fleet but not every fleet is a swarm.
“swarm is usually a fleet, but a fleet is not necessarily a swarm.”
Practical AI · 0:23:11
Chris Benson says cyberattacks spreading across systems are moving much faster in 2026 than in 2025, making it hard for human defenders to keep up.
“And it's happening very rapidly in 2026 compared to even 2025.”
Practical AI · 0:24:43
Chris Benson cites Walmart and Amazon as companies advertising drone delivery, as evidence that pervasive robotics is arriving.
“Walmart advertises drone delivery, as does Amazon, and you know, we're seeing more of those.”
Practical AI · 0:25:50
Chris Benson says he is writing a book on multi-agent architectures.
“And so I'm actually actively writing a book on this topic, so, um, among other things.”
Practical AI · 0:35:43
Chris Benson says the 'token maxing' trend of heavy AI usage faded because its cost caught up with users, shifting focus to affordability.
“it came and it went because suddenly the the cost of that caught up to people.”
Practical AI · 0:39:42
Chris Benson warns that a business tied tightly to one vendor can be badly hurt if the vendor changes its offering or absorbs a valuable function.
“all it takes is one of those vertical integrations to change something.”
Practical AI · 0:47:12
Chris Benson said China is recognizing the value of its open models, which are cheaper than expensive US proprietary models, and may restrict them.
“China is now recognizing the value it has in its open models, which are much cheaper to use than the, you know, these expensive proprietary models, you know, in the US.”
Practical AI · 0:10:17
Chris Benson said many organizations dropped fine-tuning because general models seemed good enough, but interest is now returning.
“it seemed like the models got so good that a lot of organizations said, why would we bother with that?”
Practical AI · 0:15:17
Chris Benson said he is seeing agent loops that incorporate hundreds of thousands or millions of agents.
“looking at situations where, um, loops are incorporating, um, hundreds of thousands of agents, or millions of agents”
Practical AI · 0:17:14
Chris Benson argued that future interactions between systems will occur agent to agent rather than through human-facing interfaces.
“the way interactions will occur is agentially, in a direct, you know. If you might think of it as B2B in a sense, but think of it as agent to agent.”
Practical AI · 0:23:05
Chris Benson said agentic engineering was the buzzword a few months ago and loop engineering lately, and predicted loop engineering would be antiquated before year's end.
“loop engineering, you know, we were agentic. Engineering was, you know, kind of the buzzword. Uh, a few months ago, and then lately it's loops”
Practical AI · 0:25:04
Chris Benson said he saw a quote attributed to George Lucas comparing those not keeping up with agents to people clinging to a horse and cart in the age of the automobile.
“I saw a quote earlier today from George Lucas saying, If you're not really on top of this with agentics, you're kind of holding on to the heart, the the heart, the, the, the cart, the cart and horse”
Practical AI · 0:28:19
Chris Benson described Mythos and Fable as latest-generation models able to do things no human can, with Mythos especially strong in cybersecurity.
“we're looking at things like Mythos and fable from, you know, which are the latest generation able to really, um, to really, um, you know, do things that no human is able to do, you know, Mythos, especially in terms of cybersecurity.”
Practical AI · 0:28:45
Chris Benson said Anthropic's paper describes model workspaces that serve the same function as working memory in a human brain.
“that models are creating what they're calling workspaces, where they're essentially the W. The. The notion of the workspace is fulfilling the same function as working memory in a human brain.”
Practical AI · 0:31:20
Chris Benson said Anthropic's Mythos is being used with about 150 organizations for security audits.
“this is the same organization that has Mythos out there and is working with. I believe the latest number is 150 organizations”
Practical AI · 0:03:32
Chris Benson says the report finds the AI model performance gap between the United States and China has effectively closed.
“AI model performance. Between the United States and China has closed, it's effectively closed”
Practical AI · 0:07:12
Chris Benson says Meta has moved to entirely closed models, and that it had been the leader of the US open-model contingent.
“meta, you know, with meta walking away, they're, they're now going entirely closed, and that was kind of leading the US. contingent.”
Practical AI · 0:08:24
Chris Benson says the report cites Gemini Deep Think earning an International Mathematical Olympiad gold medal while reading analog clocks accurately only part of the time.
“The example they talk about is Gemini Deep Think getting the gold medal at the IMO.”
Practical AI · 0:12:34
Chris Benson says Yann LeCun has repeatedly argued for moving past large language models toward world models.
“Yann LeCun, one of the the godfathers of AI, has talked about that many times over the last few years.”
Practical AI · 0:13:02
Chris Benson says the defense industry probably has more guardrails and responsible AI efforts than most commercial industries, because of federal regulations.
“there's probably more guardrails and responsible AI efforts around our industry than most commercial industries.”
Practical AI · 0:22:22
Chris Benson says the report ranks the US 24th in generative AI adoption, at roughly 28 percent.
“We are still, um, in the US. ranked twenty fourth at only a twenty eight, three adoption.”
Practical AI · 0:31:19
Chris Benson says Meta, the parent company of Facebook, Instagram and WhatsApp, long championed open-source models in Western countries.
“Meta, the parent company of Facebook and Instagram and WhatsApp, has long, uh, at least within Western countries, uh, the United States, long been the kind of the champion of open source models.”
Practical AI · 0:15:37
Yann LeCun, described as one of the three godfathers of AI, spent about a decade at Meta doing research.
“Yann LeCun, who is famously one of the three godfathers of AI, um, and a and a major, uh, luminary in the field, uh, had spent a decade at Meta doing research.”
Practical AI · 0:16:50
Chris Benson says part of LeCun's agreement to stay at Meta was that its models would be open sourced, which he gives as the reason the Llama family was open.
“And part of the the agreement on him staying at Meta, uh, during that time was that the models that they had would be open sourced.”
Practical AI · 0:16:50
Yann LeCun left Meta, followed by significant changes in Meta's approach to AI.
“Yann LeCun left Meta and and a lot of change ensued in terms of how Meta was approaching AI.”
Practical AI · 0:17:19
Chris Benson says Llama has increasingly trailed other frontier models in performance.
“Llama has started trailing farther and farther behind other frontier models in terms of performance.”
Practical AI · 0:17:19
Chris Benson says Meta has turned to a closed-source model family called Muse Spark, where its future effort will go.
“they have turned to a closed source model family. Now it's called Muse Spark, and that's where all of their effort will be going forward.”
Practical AI · 0:18:17
Chris Benson says China leads in open-source AI models.
“if you look at leading, uh, open source, uh, AI models, China is definitely taking the lead, uh, in that space.”
Practical AI · 0:18:50
Chris Benson says Anthropic released Claude Design about two weeks before the recording.
“two weeks ago, Claude Design came out from Anthropic and prior, you know, going into that, there were a whole a.”
Practical AI · 0:24:34
Chris Benson says OpenAI has released new image capabilities.
“OpenAI also has some im- some new image capabilities.”
Practical AI · 0:25:07
Chris Benson notes that Mythos is known about but has not been publicly released.
“Mythos came out recently and things like, Yeah, well, it's not out, but Mythos is there.”
Practical AI · 0:29:21
Chris Benson says new AI coding tools can cut work that once took months to days or weeks.
“Something that might have taken months might only take a few days or weeks at this point, uh, with the new tooling.”
Practical AI · 0:36:13
Host Chris Benson says he is a principal AI and autonomy research engineer at Lockheed Martin.
“I'm a principal, uh, AI and autonomy research engineer at Lockheed Martin.”
Practical AI · 0:00:48
Chris Benson compares today's AI job fears to the film Office Space, where two consultants called the two Bobs raised the same worries about process changes and jobs.
“they had, uh, kind of two comic characters, which maybe some folks in the audience remember, called the two Bobs, which were talking about process.”
Practical AI · 0:28:47
Chris Benson says he sponsors capstone projects at Georgia Tech's College of Computing through his nonprofit role.
“I'm a capstone sponsor for capstone projects at, uh, at Georgia Tech, uh, in the in the College of Computing”
Practical AI · 0:26:57
Chris Benson defines a financial bubble as a situation where an organization's valuation exceeds its actual value.
“the general notion of a bubble, uh, uh, you know, is it's really a financial uh concept where the valuation of an organization, uh, exceeds what its actual value is.”
Practical AI · 0:04:45