Sara Hooker Raises $50M to Challenge A.I.’s Conventional Wisdom

Sara Hooker, a prominent computer scientist, has sparked controversy within the AI industry by launching her new startup, Adaption Labs, with a $50 million investment. Hooker's ambitious project challenges conventional wisdom that more computing power leads to larger and more capable models. Instead, she believes that efficient self-learning training methods will be key to achieving significant progress in AI.

According to Hooker, the current approach of building massive models and shipping them to billions of people worldwide has reached a limit. "Most A.I. progress is: you build the biggest model, and then you ship the same model to billions of people around the world; no matter what language, what industry, what enterprise," she stated.

Hooker's startup focuses on developing models that can learn continuously and adapt to workloads in real-time as they interact with different environments. This approach would allow AI tools to respond to user feedback immediately, rather than being lost in a vacuum. Additionally, Adaption Labs is exploring alternative training methods, such as "gradient-free learning," which seeks to minimize errors without relying on optimization algorithms.

Hooker's decision to challenge the industry's conventional wisdom has caught the attention of Silicon Valley investors, who have backed her startup with significant funding. The round was led by Emergence Capital Partners, with participation from prominent venture firms like Mozilla Ventures and Alpha Intelligence Capital.

While Hooker is not the first to question traditional scaling laws in AI, she is part of a growing chorus of voices within the industry that are reevaluating their assumptions. Yann LeCun, who recently left Meta to launch AMI Labs, and David Silver, a former Google DeepMind researcher, have also raised doubts about these principles.

Adaption Labs is currently hiring for 10 roles, some of which can be based in global locations, and is offering an "Adaptive Passport" perk that allows employees to take an annual trip to a country they've never visited before. Hooker expects that the industry will soon confront the reality that ever-greater computing power is yielding diminishing returns, and algorithmic innovation will become the real driver of progress.

As Hooker puts it, "This is the year in which it will really matter."

AI's gonna be super weird if everyone starts taking smaller models to the masses

I'm low-key excited about this new startup thingy Adaption Labs ! I mean, think about it - we've been stuck on building these massive models and just shipping them out to everyone. It's like, what's the point if they're not actually learning or adapting to anything?

Sara Hooker is onto something with her idea of continuous self-learning and real-time adaptation . That sounds way more efficient and effective than just throwing a bunch of computing power at it. And I love that she's exploring new training methods like gradient-free learning .

It's about time we started questioning those traditional scaling laws in AI . We're not going to solve the world's problems with just bigger models, you know? Algorithmic innovation is where it's at!

I'm curious to see how this whole thing plays out and if Adaption Labs can really make a difference . One thing's for sure - I'll be keeping an eye on it .

I think what's interesting here is how this new approach is gonna change the way we think about AI . We've been so focused on building these massive models and shipping them out that we forgot about actually making them useful. I mean, who needs a model that's just gonna sit there not getting used? It's like buying a car that's never gonna be driven . Adaption Labs is trying to change that by focusing on something that's more practical and efficient.

It's also cool to see how this new approach aligns with some of the problems we're already facing in our daily lives, like online echo chambers and biased algorithms . If AI can learn to adapt to different environments and respond to feedback, it could be a game-changer for us all.

But what I think is even more important is that this kind of thinking is gonna take some courage . It's not just about coming up with new ideas, but also challenging the status quo and pushing people out of their comfort zones. If we want to make progress in AI (and really, in life), we need to be willing to take risks and try something new .

wonder if people realise how much we're stuck on building massive models, just 'cause that's what everyone else is doing... sounds like Sara Hooker's team might be onto something with this self-learning thingy

think about it, all those billions of people getting the same model, it's not exactly gonna help anyone, right? at least with Adaption Labs, they're trying to make AI more... human I guess, by adapting to different environments and whatnot.

also love that Silicon Valley is taking notice, but also kinda worried about all the other researchers who are just quietly working away, like Yann LeCun

I'm thinking what's up with these massive models, you know? We're already dealing with bias and misinformation, do we really need to make them bigger? I mean, Sara Hooker makes some good points about training methods and self-learning, but still, it's gonna take some time for this new approach to kick in. And $50 million is a lot of cash, she must have some solid vision

I'm curious to see how Adaption Labs will execute on this, 'gradient-free learning' sounds like some fancy tech speak But if they can make it work, that'd be a game-changer for AI. And the fact that they're offering an 'Adaptive Passport' perk? That's actually pretty cool It's like they want their employees to be part of the solution too, not just building bigger models.

It'll be interesting to see how the industry reacts to this new thinking Maybe we'll finally start to see some real progress in AI. Fingers crossed!

I'm not sure about this Adaption Labs thing... They're saying we don't need super powerful computers to make AI better? I mean, isn't that just like saying less power means less innovation? It feels like they're trying to fix a car with duct tape and hope. Efficient self-learning training methods sound cool and all, but have you seen the complexity of modern systems? I'm not convinced it's going to make that big of an impact. $50 million seems like a lot of money for a risk...

The thing about $50 million investment for Adaption Labs is that its gonna fund some serious R&D. They're exploring gradient-free learning and all that jazz . I mean, we already know that massive models aren't the answer, but this new approach could be . Did you see the chart from Yann LeCun's presentation at NYU last month? It shows how AI progress is slowing down like a zombie apocalypse . Anyway, if Sara Hooker can make it work, that'd be awesome . The industry's definitely due for a paradigm shift, and I'm curious to see where this takes us .

Average salary at Adaption Labs is $170k, btw. Not bad for a startup . They're also hiring 10 people, including some global roles . With an "Adaptive Passport" perk, employees can take an annual trip anywhere they want . The industry's gonna need more of these types of innovators to shake things up .

Current funding at Adaption Labs: $50M (investor-led)
Projected growth rate: 350% YoY

I'm loving this new direction Sara Hooker is taking AI research . This whole 'more computing power doesn't equal more capability' vibe is so true! The thought that we're basically just shipping the same old model to everyone and expecting them to magically get better results? No thanks! It's like trying to optimize a puzzle without even looking at it first .

Adaptation Labs' focus on efficient self-learning and real-time adaptations sounds like the future of AI to me . And I'm all for exploring alternative training methods that don't rely on optimization algorithms. The whole 'gradient-free learning' thing has me intrigued – who knows what kind of breakthroughs we'll see come out of this?

I think Hooker's bold move is going to shake things up in Silicon Valley and beyond . It's about time someone challenged the status quo and said, "Hey, let's rethink this whole 'more computing power equals more capability' thing!"

ooh i love this!!! Sara Hooker is literally changing the game with Adaption Labs! I mean, who needs more computing power when we can train models to learn on their own? It's about time someone dared to challenge the status quo and think outside the box. The whole " gradient-free learning" thing has me super intrigued too - can't wait to see how it plays out! Also, 10 roles up for grabs with flexible work arrangements sounds like a dream come true for so many talented folks What's even more amazing is that Hooker and the rest of these innovators are paving the way for a new era in AI. Bring it on, 2025!!!

Ugh, I'm stuck in this love-hate relationship with AI . On one hand, the idea of having more efficient self-learning training methods sounds like a total game-changer ! I mean, think about it - we're talking about an industry that's been stuck in a rut for years, building massive models and shipping them out like hotcakes . That just doesn't feel very innovative to me .

On the other hand, I'm also super worried about what this means for our society . We're already dealing with AI-generated content that's more realistic than ever before - think deepfake videos and AI-written articles that are basically indistinguishable from human-written ones . It's like we're sleepwalking into a world where the line between reality and simulation is getting super blurry .

And can I just say, $50 million isn't going to change everything ? This is just the tip of the iceberg - we need a fundamental shift in how we approach AI development . Hooker's got some great ideas, but we need more than just a fancy new startup to make a real difference .

omg i'm low-key hyped about sarah hooker's new startup , adaption labs is like, taking a major risk by challenging the status quo on ai development, but if they can pull off efficient self-learning training methods, we might be looking at a whole new era of ai! imagine being able to update models in real-time and adapt to different environments, it's like, game changing . and i love that they're focusing on user feedback too, that's something we've been missing in the industry for ages . plus, who doesn't want an "adaptive passport" perk? seriously though, sarah hooker is right, we need to rethink our assumptions about ai development and focus on algorithmic innovation, it's like, the future of tech is waiting for us .

omg I'm so down for this Sara Hooker's new startup Adaption Labs is literally shaking things up in the AI industry! I mean, who needs more computing power when you can just make your models smarter? She's on to something with that gradient-free learning thing, btw. It sounds like a total game-changer for how we approach training AI. I'm loving her emphasis on real-time adaptation and user feedback - it's so refreshing to see someone putting people first instead of just throwing more processing power at the problem.

And let's be real, who wouldn't want an "Adaptive Passport" perk? Travel is like, the ultimate stress reliever, you know? I'm rooting for Sara Hooker and her team all the way. This could be the year we start seeing some real innovation in AI, instead of just more of the same old thing. Fingers crossed, right?

I'm so stoked to see someone like Sara Hooker shaking things up in the AI industry ! I think her approach makes total sense, we're talking about AI that's actually useful and effective, not just some bloated model that's shipped out to every corner of the globe . The idea of gradient-free learning is super interesting too, it's like they're trying to optimize for progress rather than just beating the previous best result .

It's crazy to think about how much we took the whole "more computing power means more capability" thing for granted, but now people are waking up and realizing that might not be true . I love that Adaption Labs is all about creating a culture of experimentation and innovation, with perks like an adaptive passport - it sounds like they're really committed to making this work . Bring on the new era of AI, I'm excited to see what's in store!

I think she's onto something but then again maybe not... I mean, have we ever seen AI actually adapt to our needs instead of just serving up more info? It sounds too good to be true that a $50 million investment can change the game. But at the same time, how can we keep throwing billions of dollars at models that aren't even learning from us in the first place? Hooker's whole approach on gradient-free learning is kinda intriguing but wouldn't it just add more complexity to the equation? And what about those "Adaptive Passport" perks for employees? Is that just a marketing gimmick or could it actually lead to some innovative collaborations?

Ugh, I'm so over this AI startup news . Another millionaire making waves with their fancy funding rounds. Meanwhile, can we talk about how Adaption Labs' approach is basically just a rehashing of existing ideas? "Efficient self-learning training methods" sounds like buzzword marketing to me . I mean, who doesn't love the idea of gradient-free learning? It's not like this is some groundbreaking innovation .

And what really gets my goat is how Hooker is positioning herself as a rebel against the industry's conventional wisdom. Newsflash: you're not the first person to question scaling laws in AI, Yann and David were already doing that . At least they had the decency to be transparent about their methods.

I'm all for innovation, but let's not forget that Adaption Labs is just another startup trying to make a name for itself in a crowded field . And an "Adaptive Passport" perk? Please, it's just a fancy way of saying you're paying your employees to travel the world while pretending to revolutionize AI .

Sara Hooker's new startup Adaption Labs is totally shaking things up in the AI industry. I mean, think about it, we've been focusing on building these massive models for years, but honestly, who needs that much processing power? Hooker's idea of efficient self-learning training methods is a game-changer. I'm all for exploring alternative approaches like gradient-free learning. It's time to rethink our assumptions about what drives progress in AI.

I love how Adaption Labs is prioritizing adaptability and user feedback over just churning out more models. That "Adaptive Passport" perk? Genius! If the industry starts to realize that bigger isn't always better, we might actually start seeing some real innovation happen. This year could be a major turning point for AI - I'm excited to see what's in store!

I'm intrigued by Sara Hooker's approach to AI development . While I agree that the current method of building massive models and shipping them to everyone might be a limiting factor, I'm not entirely convinced that efficient self-learning training methods are the key to progress .

I mean, isn't there still a lot we don't understand about human behavior and cognition? Are we really ready to abandon traditional scaling laws for AI just yet? And what's this "gradient-free learning" she's talking about? Sounds like some fancy math stuff that might work in theory but is hard to put into practice .

Still, I gotta give Hooker props for thinking outside the box and challenging the status quo . The idea of an "Adaptive Passport" perk is pretty cool too . Maybe she's onto something here... or maybe not . Time will tell if her approach will revolutionize the AI industry or just be another flash in the pan

I don't know if I'm for or against this whole thing... Like, on one hand, I get what Sara Hooker is trying to do - we need AI that's more adaptable and responsive, you feel? But then again, isn't she also kinda dismissing the idea of bigger models being better just because they're not shipping them to billions of people worldwide? IDK, maybe she's onto something? Maybe we've been doing this whole "build massive model" thing for too long. On the other hand, what if this new approach is all flash and no substance?

I'm intrigued by this new startup from Sara Hooker . I think she's onto something with her approach to AI development. The idea that efficient self-learning training methods are key to making progress in AI resonates with me. It makes sense that just building massive models and shipping them out won't be enough anymore .

I'm also curious about this "gradient-free learning" method she's exploring. If it can really minimize errors without relying on optimization algorithms, that could be a game-changer for the industry . And I love that Adaption Labs is prioritizing flexibility and adaptability in their development process - it's not just about throwing more computing power at the problem, but also about creating models that can learn and respond in real-time .

The timing of this startup couldn't be better, either. With all the hype around AI and machine learning, it's refreshing to see someone questioning the conventional wisdom . I'm excited to see how Adaption Labs evolves and what kind of impact they can make in the industry .