Despite growing chatter about a future when much human work is automated by AI, one of the ironies of this current tech boom is how stubbornly reliant on human beings it remains, specifically the process of training AI models using reinforcement learning from human feedback (RLHF).

At its simplest, RLHF is a tutoring system: after an AI is trained on curated data, it still makes mistakes or sounds robotic. Human contractors are then hired en masse by AI labs to rate and rank a new model’s outputs while it trains, and the model learns from their ratings, adjusting its behavior to offer higher-rated outputs. This process is all the more important as AI expands to produce multimedia outputs like video, audio, and imagery which may have more nuanced and subjective measures of quality.

Historically, this tutoring process has been a massive logistical headache and PR nightmare for AI companies, relying on fragmented networks of foreign contractors and static labeling pools in specific, low-income geographic hubs, cast by the media as low wage — even exploitative. It’s also inefficient: requiring AI labs wait weeks or months for a single batch of feedback, delaying model progress.

Now a new startup has emerged to make the process far more efficient: Rapidata‘s platform effectively “gamifies” RLHF by pushing said review tasks around the globe to nearly 20 million users of popular apps, including Duolingo or Candy Crush, in the form of short, opt-in review tasks they can choose to complete in place of watching mobile ads, with data sent back to a commissioning AI lab instantly.

As shared with VentureBeat in a press release, this platform allows AI labs to “iterate on models in near-real-time,” significantly shortening development timelines compared to traditional methods.

CEO and founder Jason Corkill stated in the same release that Rapidata makes “human judgment available at a global scale and near real time, unlocking a future where AI teams can run constant feedback loops and build systems that evolve every day instead of every release cycle.””

Rapidata treats RLHF as high-speed infrastructure rather than a manual labor problem. Today, the company exclusively announced to us at VentureBeat its emergence with an $8.5 million seed round co-led by Canaan Partners and IA Ventures, with participation from Acequia Capital and BlueYard, to scale its unique approach to on-demand human data.

The pub conversation that built a human cloud

The genesis of Rapidata was born not in a boardroom, but at a table over a few beers. When Corkill was a student at ETH Zurich, working in robotics and computer vision, when he hit the wall that every AI engineer eventually faces: the data annotation bottleneck.

“Specifically, I’ve been working in robotics, AI and computer vision for quite a few years now, studied at ETH here in Zurich, and just always was frustrated with data annotation,” Corkill recalled in a recent interview. “Always when you needed humans or human data annotation, that’s kind of when your project was stopped in its tracks, because up until then, you could move it forward by just pushing longer nights. But when you needed the large scale human annotation, you had to go to someone and then wait for a few weeks”.

Frustrated by this delay, Corkill and his co-founders realized that the existing labor model for AI was fundamentally broken for a world moving at the speed of modern compute. While compute scales exponentially, the traditional human workforce—bound by manual onboarding, regional hiring, and slow payment cycles—does not. Rapidata was born from the idea that human judgment could be delivered as a globally distributed, near-instantaneous service.

Technology: Turning digital footprints into training data

The core innovation of Rapidata lies in its distribution method. Rather than hiring full-time annotators in specific regions, Rapidata leverages the existing attention economy of the mobile app world. By partnering with third-party apps like Candy Crush or Duolingo, Rapidata offers users a choice: watch a traditional ad or spend a few seconds providing feedback for an AI model.

“The users are asked, ‘Hey, would you rather instead of watching ads and having, you know, companies buy your eyeballs like that, would you rather like annotate some data, give feedback?’” Corkill explained. According to Corkill, between 50% and 60% of users opt for the feedback task over a traditional video advertisement.

This “crowd intelligence” approach allows AI teams to tap into a diverse, global demographic at an unprecedented scale.

• The global network: Rapidata currently reaches between 15 and 20 million people.

• Massive parallelism: The platform can process 1.5 million human annotations in a single hour.

• Speed: Feedback cycles that previously took weeks or months are reduced to hours or even minutes.

• Quality control: The platform builds trust and expertise profiles for respondents over time, ensuring that complex questions are matched with the most relevant human judges.

• Anonymity: While users are tracked via anonymized IDs to ensure consistency and reliability, Rapidata does not collect personal identities, maintaining privacy while optimizing for data quality.

Online RLHF: Moving into the GPU

The most significant technological leap Rapidata is enabling is what Corkill describes as “online RLHF”. Traditionally, AI is trained in disconnected batches: you train the model, stop, send data to humans, wait weeks for labels, and then resume. This creates a “circle” of information that often lacks fresh human input.

Rapidata is moving this judgment directly into the training loop. Because their network is so fast, they can integrate via API directly with the GPUs running the model.

“We’ve always had this idea of reinforcement learning for human feedback… so far, you always had to do it like in batches,” Corkill said. “Now, if you go all the way down, we have a few clients now where, because we’re so fast, we can be directly, basically in the process, like in in the processor on the GPU right, and the GPU calculate some output, and it can immediately request from us in a distributed fashion. ‘Oh, I need, I need, I need a human to look at this.’ I get the answer and then apply that loss, which has not been possible so far”.

Currently, the platform supports roughly 5,500 humans per minute providing live feedback to models running on thousands of GPUs. This prevents “reward model hacking,” where two AI models trick each other in a feedback loop, by grounding the training in actual human nuance.

Product: Solving for taste and global context

As AI moves beyond simple object recognition into generative media, the requirements for data labeling have evolved from objective tagging to subjective “taste-based” curation. It is no longer just about “is this a cat?” but rather “is this voice synthesis convincing?” or “which of these two summaries feels more professional?”.

Lily Clifford, CEO of the voice AI startup Rime, notes that Rapidata has been transformative for testing models in real-world contexts. “Previously, gathering meaningful feedback meant cobbling together vendors and surveys, segment by segment, or country by country, which didn’t scale,” Clifford said. Using Rapidata, Rime can reach the right audiences—whether in Sweden, Serbia, or the United States—and see how models perform in real customer workflows in days, not months.

“Most models are factually correct, but I’m sure you’re you have received emails that feel, you know, not authentic, right?” Corkill noted. “You can smell an AI email, you can smell an AI image or a video, it’s immediately clear to you… these models still don’t feel human, and you need human feedback to do that”.

The economic and operational shift

From an operational standpoint, Rapidata positions itself as an infrastructure layer that eliminates the need for companies to manage their own custom annotation operations. By providing a scalable network, the company is lowering the barrier to entry for AI teams that previously struggled with the cost and complexity of traditional feedback loops.

Jared Newman of Canaan Partners, who led the investment, suggests that this infrastructure is essential for the next generation of AI. “Every serious AI deployment depends on human judgment somewhere in the lifecycle,” Newman said. “As models move from expertise-based tasks to taste-based curation, the demand for scalable human feedback will grow dramatically”.

A future of human use

While the current focus is on the model labs of the Bay Area, Corkill sees a future where the AI models themselves become the primary customers of human judgment. He calls this “human use”.

In this vision, a car designer AI wouldn’t just generate a generic vehicle; it could programmatically call Rapidata to ask 25,000 people in the French market what they think of a specific aesthetic, iterate on that feedback, and refine its design within hours.

“Society is in constant flux,” Corkill noted, addressing the trend of using AI to simulate human behavior. “If they simulate a society now, the simulation will be stable for and maybe mirror ours for a few months, but then it completely changes, because society has changed and has developed completely differently”.

By creating a distributed, programmatic way to access human brain capacity worldwide, Rapidata is positioning itself as the vital interconnect between silicon and society. With $8.5 million in new funding, the company plans to move aggressively to ensure that as AI scales, the human element is no longer a bottleneck, but a real-time feature.

The company’s customised AI chips aim to achieve cheaper and faster results than traditional AI hardware.

Toronto-based start-up Taalas has raised $169m for its specialised AI hardware models.

Total investment in the company stands at $219m, with funding from Quiet Capital and Fidelity, among others, according to Reuters.

In a blogpost from CEO Ljubisa Bajic announcing the release of its first models, the company said it wants to mitigate the “high latency and astronomical cost” of AI, and that its specialised method is faster and cheaper than traditional AI chip approaches.

The company said it took a team of 24 and a spend of $30m since its founding less than three years ago to bring to market its first product, a hard-wired Llama 3.1 8B, which is available as both a chatbot demo and an inference API service.

The company’s aim is to mitigate the need for vast and expensive data centres through the principles of specialisation, merging storage with computation, and simplification.

Taalas said its “platform for transforming any AI model into custom silicon” means that “from the moment a previously unseen model is received, it can be realised in hardware in only two months”.

It claimed its hardware output is “an order of magnitude faster, cheaper and lower power than software-based implementations”, achieved through physically customising chips depending on the bespoke needs of the AI model in question.

Taalas claimed its silicon Llama chip, for example, is nearly 10 times faster than the current state of the art, costs 20 times less to build and consumes 10 times less power.

Taalas aims to release two further models in 2026.

AI chipmaking giant Nvidia this week announced a huge deal with Meta to provide millions of chips for Meta’s AI infrastructure in exchange for billions of dollars.

Don’t miss out on the knowledge you need to succeed. Sign up for the Daily Brief, Silicon Republic’s digest of need-to-know sci-tech news.

The Nothing Phone (4a) series is shaping up to be more expensive than its predecessor according to fresh leaks detailing pricing, specs and release dates ahead of the company’s March 5 launch event.

A new report from Dealabs suggests the Nothing Phone (4a) will start at around €400. This marks roughly a €50 increase over the Phone (3a), with pricing said to vary slightly by region. For instance, Germany and Spain will reportedly see a €389 starting price. Meanwhile, France, Belgium and Italy could see it land at €409. A 12GB RAM variant is expected to cost between €429 and €449.

The Nothing Phone (4a) Pro could see an even steeper jump. Dealabs claims pricing will begin at €479 in Germany and Spain. This rises to €499 in France, Belgium and Italy — around €90 more than the previous Pro model. A higher-tier 12GB version could reach as much as €569 depending on the market.

It has never been more expensive to buy a new car. The average transaction price last month for buyers in the United States was $48,576, up nearly a third from 2019, according to Edmunds. The “affordable” car—$20,000 or less—is dead.

The high prices have been pinned on plenty of economic dynamics: lingering pandemic-era supply-chain issues, the introduction of expensive technology into everyday cars, higher labor and raw materials costs, and new tariffs by the Trump administration affecting imported steel, aluminum, and cars themselves.

Now, despite a US Supreme Court ruling that will nix some of those Trump tariffs, car buyers will likely get no respite.

“The core cost structure facing the auto industry hasn’t fundamentally changed overnight,” writes Jessica Caldwell, Edmunds’ head of insights, in an emailed statement. Put more simply: Cheaper cars aren’t coming, at least not because of this ruling.

The Supreme Court’s decision gets in the way of the president’s power to use the International Emergency Economic Power Act to levy tariffs in response to emergencies. Trump used this power to apply tariffs to countries around the globe, the emergency being “large and persistent” trade deficits. The administration applied other new duties on Canada, China, and Mexico because of what it called emergencies related to the flow of migrants and drugs into the United States.

But most of the tariffs that affect the auto industry come from another law, section 232 of the Trade Expansion Act. That provision can apply to imports that “threaten to impair” the country’s national security. Tariffs on steel, aluminum, copper—key raw materials for cars—and imported auto parts and vehicles themselves came under this provision and are still in effect. This includes 15 percent tariffs on cars built in Europe, Japan, and South Korea.

Automakers have actually done an OK job shielding consumers from the effects of tariffs, Caldwell says. Even as retailers have blamed tariffs for steadily rising prices of consumer goods like electronics and appliances, car prices are up just 1 percent since this time last year, the firm’s data shows. But as the tariff regime drags on, that could change in ways that make new-car buyers even less happy.

“If cost pressures continue to build, automakers may have less room to shield shoppers from higher prices,” Caldwell says, “but for now, the broader market impact is still playing out.”

Reportedly, the organisation intends to hire up to 100 additional employees within its Seed artificial intelligence division.

According to Bloomberg, Chinese technology giant ByteDance has announced plans to employ up to 100 new people in its artificial intelligence (AI) division as a means of competing with some of the world’s leading US-based AI companies. 

The positions open to US-based professionals will be in Seed, which is ByteDance’s AI division with laboratories in the US, Singapore and China. 

Vacancies will be across various responsibilities and will include work in producing international data for ByteDance’s large language models, advancing its popular text, image and video generation tools, completing research to develop ‘human-like’ AI, and building science models that enable the organisation to pursue drug discovery and design. 

ByteDance’s announcement comes despite significant concerns from US lawmakers and regulators around national security. Those in the regulatory space in the US have long argued that there is potential for ByteDance to use TikTok in the collection of its citizens’ private, valuable data or in the spreading of a narrative in favour of Beijing’s leadership via the app’s algorithm. 

Last month, after a long period of deliberation, stalling and false starts, ByteDance and representatives in the US reached a deal wherein the organisation would create an entity for US TikTok operations – with non-Chinese majority owners – as a means of addressing some of the pressing security concerns. 

As noted by Bloomberg, many know ByteDance solely in the context of popular social media platform TikTok; however, it also operates as a well-known AI company in China, offering chatbot app Doubao, AI video generation model Seedance 2.0, and image generation mode, Seedream 5.0.

Earlier this week (16 February), ByteDance promised to “strengthen current safeguards” against intellectual property theft after Disney threatened legal action regarding videos generated by Seedance 2.0.

Via a cease and desist letter, Disney claimed that Seedance 2.0 operates a “pirated library” of Disney assets, taken from its biggest franchises. The company accused ByteDance of using its proprietary content assets as if they were in the public domain.

Don’t miss out on the knowledge you need to succeed. Sign up for the Daily Brief, Silicon Republic’s digest of need-to-know sci-tech news.

In a year defined by sovereign AI and data, one truth has become unavoidable: humans will matter more than ever. Even the most ambitious AI strategies will stall if organizations fail to invest in their people.

More than 95% of enterprises worldwide now say they want to operate as their own AI and data platforms within the next 850 working days. It’s a stunning recognition from C-suite leaders across 13 countries representing a combined GDP of $48 trillion – and a signal of how rapidly the world is shifting. IDC estimates this transition could generate $17 trillion in GDP growth, effectively creating the world’s third-largest economy if counted as a country.

Yet despite this massive ambition, only 13% of more than 134,000 major enterprises are getting it right.

These early leaders have made AI and data sovereignty a mission-critical priority. Their infrastructure allows intelligence to be accessed securely – anywhere, anytime, and in any form. The results speak for themselves: they see 5x higher ROI than the rest, with 2x more GenAI and agentic systems deployed in mainstream production. They are also 250% more confident in their ability to thrive long term.

Enterprises such as Abbott, AIA Singapore, Aviva India, Boston Scientific, Danske Bank, ENOC, JP Morgan Chase, Mastercard, Singtel, Wells Fargo, Toyota, and others are already proving what scaled success looks like.

But this transformation is not a push-button upgrade. Digital transformation took nearly a decade. The AI and data revolution may peak in just three to four years, and its impact could far surpass anything seen before.

That is why the defining question of the next era is not purely technological. Sovereign AI will rise or fall on human readiness. Organizations that fail to reskill, align, and carry their workforce into this transformation will find their ambitions constrained before they ever scale.

There are three key reasons why.

The Intelligent Systems Economy Will Require Hundreds of Millions of Skilled People

This new AI-driven economy brings greater complexity than the cloud migration wave. According to the World Economic Forum’s 2025 Future of Jobs report, AI is expected to displace 92 million jobs, but also create 170 million new ones—a net gain of 78 million. In some countries, up to 70% of these new roles risk going unfilled due to skill shortages.

“We cannot realize the potential of this new intelligent systems economy unless we invest significant time and energy reskilling and enabling employees in new ways,” says Einav Lavi, CHRO at EDB. “Demand for qualified people will far exceed supply, emphasizing how central humans are to this revolution.”

Enterprise-Wide Agentic Success Requires Everyone – Not Just Specialists

The top 13% of enterprises treat AI and data sovereignty as a company-wide standard. Their 2x density of AI initiatives and 5x ROI stem from building a sovereign foundation that reaches everyone—from HR and frontline staff to product design, engineering, and finance.

They rolled out GenAI and agentic systems in a coordinated, enterprise-wide sequence that embedded AI into organizational DNA. Skill levels varied, but reskilling at scale produced transformation at scale.

As enterprises evolve into AI “factories,” every employee becomes part of the production line, sharing common standards, practices, and a unified vision.

The Future Workforce Will Demand Continuous Reinvention

For most of the past century, people held 1.5 careers across 5–10 employers. That era is ending. By 2050, 60–80% of today’s jobs will be automated, and individuals may have 20–30 roles across a dozen organizations.

“In this environment, continuous reskilling becomes one of the most valuable currencies of success,” Lavi notes. “The enterprises that thrive will invest as much in their people as they do in their AI.”

AI itself will accelerate this reinvention – surfacing internal opportunities faster, matching people to roles or stretch assignments, and building personalized development pathways. Growth will be driven increasingly by skills and contribution, not proximity or bias.

For HR and managers, AI-powered “people copilots” will reshape workforce planning by identifying early signals of burnout, workload imbalance, sentiment shifts, and retention risks – augmenting, not replacing, human judgment.

The goal is not the automation of humanity, but the elevation of what makes us human – freeing people to focus on creativity, judgment, empathy, and innovation, the very things machines cannot replicate.

Digital Trends partners with external contributors. All contributor content is reviewed by the Digital Trends editorial staff.