CULTURE

How AI changes what it means to be a filmmaker

On Thursday, I sat in a small theater in Santa Monica among hundreds of other viewers and had the distinct realization that Hollywood doesn't know what's about to hit it. 

The Deep View was invited to Runway's fourth annual AI Festival, showing off the ten best AI-generated films from hundreds of submissions. The short films spanned a wide range, some animated, some stop-motion, some realistic (because I hesitate to use the phrase "live action"), ranging from body horror to emotionally pensive to just plain outlandish.

There were, of course, some of the usual giveaways. Mouth movements that didn't line up quite right, over-smoothness of skin and motion, and a consistent inability to generate legible text. But walking out of the theater, I realized just how far we've come from Will Smith eating spaghetti. 

"It's just been incredible to see how fantastic the tools I've gotten that actually are usable," Dave Clark, cofounder of AI film studio Promise and creator of Tairell Isn't Real, one of the films shown at the festival, told The Deep View. "Now I can actually cut in a 4K generative AI shot against something I shoot on a camera." 

Hollywood has long been at war with itself over AI. Many artists and creatives have fiercely rallied against it, and major awards institutions are navigating how the tech affects eligibility for the industry's most prestigious honors. Some are trying to come up with solutions, such as the "Human Consent Standard," to give actors more control over how their likeness is used by AI. 

However, AI's use in entertainment has gone far beyond fringe technoevangelists. Runway, for instance, has a partnership with Lionsgate, which the companies expanded last week to create a joint development program to roll out a "slate of co-developed projects blending AI and content." Other high-profile Hollywood voices have also taken an interest in the tech, including Martin Scorsese's highly contentious decision to become an adviser to Black Forest Labs. 

The argument often made by companies like Runway is that AI is meant to raise the floor. Those that didn't have access to the tools needed to manifest their ideas now have the ability to create something out of nothing, Jamie Umpherson, chief creative officer at Runway, told The Deep View. Meanwhile, those who are already established in their careers have an entirely new arsenal of tools at their disposal. 

"It's really opening the door to a lot more creatives and filmmakers to get their first chance to get a project off the ground that might have been sitting there for 10 years," said Umpherson. 

And despite the fears that AI will put creatives out of work, in a panel with Runway CEO Cristobal Valenzuela before the screening, Roger Avary, award-winning screenwriter and partner in the production company General Cinema Dynamics, said, "The only people who are really going to lose their jobs are the gatekeepers."

"I think we're seeing a resurgence right now in the cinemas of indie filmmaking, and I think using AI tools ... people that come together to make something that is just the glimmer in their eye," said Gala Avary, a partner in General Cinema Dynamics, on the panel. "Suddenly, things that weren't possible, like locations and permitting, all of a sudden you're able to do those things with these AI tools."

As major Hollywood studios take fewer risks, smaller indie films are gaining traction among audiences, as evidenced by the massive popularity of films like "Obsession" and "Backrooms" compared to big-budget, blockbuster Disney and Amazon movies that came out around the same time. It's entirely possible that, in the best-case scenario, AI will enable more of those ideas to come to life. At the same time, money is money, and if these major studios can save some of it by replacing creatives with AI, they will likely do so. Still, Runway's AI film festival left me brimming with even bigger questions than the economics of it all, wondering, as a creative myself, what it means to be an artist when a machine can do the creating for you. The biggest of them: If AI tools make everyone good from the jump, will virtuosity become obsolete? Is it still important to be bad at something in order to appreciate how hard it is to get good at something?

TOGETHER WITH IBM

Guiding AI with project-level rules

AI agents are reshaping development, reportedly generating an average of 48% of code for surveyed organizations. But with that shift comes growing concern: 55% of engineering leaders surveyed are concerned about losing shared understanding of how their codebase evolves, and 39% are worried about shipping with confidence. 

The issue is not that AI produces bad code. It is that each agent makes different decisions on frameworks, testing and patterns. Over time, this can create inconsistency that’s harder to review and maintain. Project-level rules provide a structured way to address this, encoding conventions and standards directly into workflows so AI-generated code remains aligned with how teams build and maintain software.

See how to help keep AI code consistent

WORKFORCE

AI impact on jobs is overstated, says IBM CHRO

Workers are anxious that AI is coming for their jobs. But some HR leaders say those fears may be unwarranted.

AI will eliminate relatively few roles overall, according to Nickle LaMoreaux, chief human resources officer at IBM. While highly administrative and process-driven jobs are likely to disappear, she says the productivity gains from AI will create entirely new roles that don't yet exist.

"As economies grow, we're going to need more jobs," LaMoreaux told The Deep View at New York City Tech Week earlier this month. "AI will take a portion of that, but because we're growing, we still need human jobs to fill that other space."

It's too early to know exactly what those jobs will look like. Instead of focusing on which roles AI might replace, LaMoreaux says companies should think about how employees can use the time freed up by automation. 

That optimistic outlook may offer little comfort to early-career workers.

"If you now have the capacity freed up, can you deploy those employees into new models, new customer needs, new products, or honestly just a long list of work you never get to in your organization?" LaMoreaux asks.

Entry-level roles exposed to AI, including customer service, data entry, and coding, are in "substantial decline," according to recent research from Stanford University's Digital Economy Lab. The study found that knowledge workers with college degrees could face some of the biggest disruptions.

Even so, LaMoreaux says recent graduates can take steps to stand out. Her advice: focus less on responsibilities and more on outcomes. Whether you managed a budget for a volunteer organization or published a research paper, be prepared to explain what you accomplished, how you achieved it, and the impact it had. That principle also applies to AI literacy, a skill employers increasingly want to see in candidates.

Highlighting how you used AI to solve a specific problem, such as building an agent to automate a task, is more compelling than simply listing proficiency with ChatGPT on a résumé.

"AI literacy is important, but I think it's less about knowing a specific tool and more about your application of it," LaMoreaux says. 

Regardless of industry or job title, soft skills will remain difficult to automate. Communication, leadership, judgment, and conflict management, LaMoreaux says, "will not be taken over" by AI.

AI's impact on jobs is becoming a growing source of anxiety for workers. As companies lay off workers and cite AI as a way to cut costs, employees are left trying to figure out what comes next. Some workers are changing careers, while college students are switching majors in search of jobs they believe are less vulnerable to automation. But many of these decisions are being made before AI's long-term impact on work is fully understood. History suggests technological shifts create new jobs. However, AI raises a different question. If AI can perform most of a job's tasks, what work is left for humans? The real challenge may not be whether AI creates new jobs, but whether it creates the kinds of jobs people actually want.

TOGETHER WITH CDATA

Single-source MCP servers fail 1 in 4 prompts

Enterprise AI failure patterns are consistent: schema mapping, date logic, multi-filter conditions, and write validations.  

Without a governed context layer, integrating single-source MCP servers becomes a nonstarter at scale for the exact operations enterprise workflows depend on. 

CData Connect AI resolves all four failure patterns  with governance built into the architecture: role-based access control, per-user authentication, downscoped toolsets, and query-level audit trails.  

See how Connect AI handles what single-source servers can't, now offering 17 new sources including GitLab, Gong, BambooHR, and Zoom, on top of hundreds already supported. 

Start connecting for free ->

CONSUMER

Why AR's comeback now runs through AI

A decade ago, Google Glass, Microsoft HoloLens, and Magic Leap made augmented reality (AR) devices look inevitable. Then Meta bet its future on the metaverse. Neither paid off, but AI may be the catalyst AR needed.

The appeal of AR is simple: wear a device that layers graphics like messages, directions, and information over your environment, no other device needed. At a time when people are actively trying to spend less time on their devices and out in the real world, this should have been a no-brainer. So why hasn't it taken off? It comes down to three factors: utility, form factor, and price.

Convincing people that they can ditch their laptop or phone for something to wear on their face takes a lot of mental reframing. That is especially difficult to do when these devices are often bulky, heavy headsets that, in most cases, require another dangling accessory to be tethered to, such as a computing puck or a battery. Even for the most experimental of users, paying to play can be a bit difficult as the price tags are high, with headsets like the Apple Vision Pro retailing for $3,500. 

But to fix all three of these issues, the first step may be to convince people of the utility of AI. 

There has been an increase in demand for smart glasses in the past year, influenced by people wanting an AI assistant to break out of just a chatbot interface on a device screen and into the real world. According to IDC data, smart glasses without displays surged 167% year-over-year in the first quarter of 2026, with approximately 2.25 million units sold. That has trickled into AR, ER, MR, and VR categories, which the same data says have grown 86% year-over-year.

While smart glasses without displays have been shown to provide AI with more context via cameras and microphones, AR can take this much further. Not only can you give the device context about the world through its cameras, but you can also see the results overlaid directly in your environment. For instance, asking something like 'How would this piece of furniture look in my living room?' would show you the answer anchored in your actual space, in real-life proportions.

The rise of AI has also pushed manufacturers to develop more capable wearable processors, like the Snapdragon AR Elite that Qualcomm unveiled at AWE. As those chips get smaller and more powerful, so too could the headsets that house them, directly tackling the form factor problem.

We saw a glimpse of this at AWE with both the XREAL Aura and the new Snap Specs, which mimic the form factor of regular glasses, albeit chunkier, ditching the traditional headset entirely for AR experiences.

As demand grows and companies iterate, prices for AR glasses will naturally drop, and the hardware will improve. That will mean longer battery life, lighter frames, and all-day wearability, potentially giving AR the mainstream moment it has been waiting for. The Snap Specs offer the biggest meaningful step toward that reality, but it's still several steps away from being a product anyone will buy to replace their phone. Still, the path ahead has never been clearer, thanks to breakthroughs in chips, displays, and general miniaturization of technologies. And, AI is likely to continue to fuel new features to make smart glasses more useful and hardware advances that will accelerate progress toward AR.

LINKS

• Europe 2031, a speculative experiment, predicts EU instability in AI race

• Google DeepMind's John Jumper departs for Anthropic after nine years

• Norway sets a near ban on generative AI use in elementary schools 

• Amazon MGM drops movie on Sam Altman after OpenAI partnership

• Barret Zoph is leaving OpenAI again after five months back

• Anduril explores establishing operations in Israel

• Claude Design: Anthropic's design tool now remembers your context and stays on-brand across projects.  

• ChatGPT: OpenAI's flagship chatbot updated the way users take and upload photos on iOS. 

• Cohere North Mini Code: In an update, the company released 4-bit quant, its first open-source agentic coding model small enough to be run on a Mac. 

• Codex: OpenAI's coding platform can now hand off project threads between local and remote hosts.

• OpenAI: Researcher, Trustworthy AI

• Anthropic: Research Engineer, Safeguards Labs

• Salesforce: Adversarial AI & Research Engineer

• Google DeepMind: Research Engineer, Security and Privacy

GAMES