Is an era of human-centric A.I. on the way?
A hot new startup is pitching itself with a human-first mission in the automation boom. The devil will be in the details.
Film-lovers and festival-goers: Hard Reset is hosting the panel discussion “Streaming Power, Creative Loss: The Fight for Hollywood’s Future” at Sundance this coming Monday featuring former HBO Films executive Keri Putnam, documentary producer Shane Boris, and antitrust lawyer Lee Hepner. Moderated by Richard Rushfield from The Ankler. You can register here.
A story this week caught my eye: The New York Times profile of Humans&, a company that has “embraced the notion that A.I. should empower people rather than replace them.”
The story and the prominent placement it was given — not to mention the $480 million in seed funding it has raised — seems a sign that questions about how A.I. can augment but not replace human labor seem to be growing.
“A lot of our investors are human, and they care where humanity is going,” Georges Harik, one of the startup’s founders, who was Google’s seventh employee, told the Times. I wonder who the other investors are. Other members of the leadership come from Anthropic and xAI. The company’s website says that “for humans, progress happens when we understand one another, build trust, make connections, and work together. That is where we believe the next chapter of AI should begin.” But it is light on specifics.
It is a sign the culture is shifting? A smart marketing pitch that plays to public concerns about A.I? Or just another marker of the frothiness in A.I. startup land? All maybes.
The company is just three months old and is already valued at a potential $4.8 billion, with investments from big names like Nvidia, Jeff Bezos, and Laurene Powell Jobs’ Emerson Collective. This level of eye-popping funding is now a fairly quotidian aspect of the A.I. rush, as TechCrunch notes. Thinking Machines Lab, founded by former OpenAI CTO Mira Murati, raised a whopping $2 billion last July. (Meanwhile, some of the company’s founding team departed recently.) Unconventional AI, which is seeking to build a more energy-efficient computing system for A.I., raised $475 million in December.
And a few years now into the A.I. boom, questions about how to augment vs. replace humans, complicated and fraught as ever, seem to be picking up. Particularly when a measure of research over the last year has shown the limits A.I. systems are bumping up against when they are employed on the job. Yes, as A.I. believers note, the systems are improving rapidly and will continue to. But more human-centric design could potentially make for a break with the culture of anthropomorphized chatbots — and the underground economy of credible fakes — that have been a part of the first few years of A.I. growth.
The Times traces the lineage of these ideas to the Institute for Human-Centered AI, a lab founded at Stanford in 2019.
“There are some pretty clear principles that designers are beginning to take that favor human-centered approaches,” Ben Shneiderman, a professor at the University of Maryland who runs an online community about the issue, told the Times. “It goes back to the fundamental principle of ensuring human control.”
And now, a brief personal note:
This is my last post for Hard Reset! I’ve accepted an offer for a full-time reporter position here in San Francisco where I’ll have more time to do what I love: report and write in-depth stories. More on that later. This newsletter has been a gratifying and fun project these last 12 months and I’m looking forward to watching it continue to grow under the sharp penmanship of Ariella, Alex, Jake and others. I’ve enjoyed learning about Substack, a dynamic ecosystem with a lot of worthwhile writing and intellectual exploration. Some of my favorite stories from my time here include pieces with A.I. researcher Damien Charlotin and the chillingly prescient Jack A. Goldstone. I am very grateful to all the sources who have worked with me, and for you — our readers. Please give me a holler on your social site of choice: LinkedIn, X, BlueSky, or Signal (@elirosenberg.30), and let’s stay in touch!
Here’s what else I’m reading this week:
Do you ever find yourself in the position where a big news story sneaks by you — the headlines speed by, but don’t draw you in? Eventually, the story becomes impossible to ignore, but it’s too late. The saga of the Ellisons, Paramount, and their attempt to buy Warner Bros. Discovery has been one of those stories for me. But this big takeout in New York magazine, a very well-reported profile of David Ellison, comes to the rescue. It’s a great way to catch-up on the saga and understand the bigger forces at play — a fascinating parable about what happens as tech money makes inroads in Hollywood.
For those interested in the newsletter boom, Beehiiv is on the rise, as some folks leave Substack over concerns about the cut they take from writers, the dynamics and algorithms that increasingly influence the platform, and the hands off approach it takes to moderating the content on its site.
Beehiiv now draws more than 40,000 monthly active users, including nearly 15,000 paying subscribers, with roughly one in seven new writers arriving from Substack… (Reuters)
Portugal has ordered Polymarket, which lets users place bets on current events, to be blocked in the country after wagers surged before the results of a recent election were announced. Cryptocurrency-focused news site The Block notes a growing list of countries moving against the platform in Europe, including France, Germany and Hungary. Meanwhile, trading/betting volume on Polymarket and rival site Kalshi has grown sharply this year.
An NLRB complaint accuses architecture firm Snohetta of illegally dismissing eight employees after a failed union drive among the staff in 2023, the Times reports. Elaine Molinar, a partner at the firm, told the Times that “All decisions regarding the work force reduction were driven by business considerations that started long before the unionization effort.”
I thought this Casey Newton post about coding his own tools for writing and research was interesting.
And, on the topic of Greenland, whose rare earth minerals may or may not be critical to the technological arms race with China, depending on who you ask:
The reality is that Denmark’s ownership of Greenland is not what’s stopping the United States from tapping the island’s treasure trove [of minerals]. It’s the punishing Arctic environment.
Researchers say it would be extremely difficult and expensive to extract Greenland’s minerals because many of the island’s mineral deposits are located in remote areas above the Arctic Circle, where there is a mile-thick polar ice sheet and darkness reigns much of the year. (CNN)
Thanks y’all, it’s been a blast.




The timing of Humans& raising half a billion with this pitch feels intentional. Companies are realizing they oversold automation potential and now there's fatigue setting in from employees who've spent two years wondering if their job's next. Had a manager last month tell me they're walking back some AI tools cause the team just stopped trusting internal processes. Smart play to frame AI as augmentation when the replacement narrative is causing more friction than adoption.