- The Deep View
- Posts
- Human oversight remains crucial as AI testing surges
Human oversight remains crucial as AI testing surges

Welcome back! The US-UK Technology Partnership is driving a wave of AI news this week, including Microsoft signing a $6.2 billion, five-year agreement with Nscale Global Holdings Ltd. and Aker ASA to build European-owned hyperscale AI infrastructure in Norway, a deal that secures Microsoft's access to advanced AI compute capacity powered entirely by renewable energy.
1. Human oversight still key as AI in testing soars, report
2. Figure AI inches closer to mass deploying humanoid robots
3. Invisible raises $100M, now worth $2B
AI AGENTS
Human oversight still key as AI in testing soars, report

Despite the rocketing use of AI in testing digital products, human input remains hugely important. That’s the key finding of the State of Digital Quality in Functional Testing in 2025 report.
Now in its fourth year, the annual report is compiled by Applause, a company specializing in digital quality and outsourced testing, and is designed to help organizations and companies achieve the highest standards for their websites and apps.
In the 12 months since the last report, the use of AI in testing has doubled, with 60% of respondents deploying generative and agentic solutions, as opposed to 30% in 2024. Digging a little deeper into the figures reveals that the most common uses involve developing test cases (70%), optimizing testing (55%) and automating test scripts (48%).
However, while the rate of adoption is accelerating at breakneck speed, this presents a challenge for many organizations, with 80% admitting a lack of in-house testing expertise hinders them.
Among the specific areas of concern is the need to keep up with rapidly changing requirements (cited by 92%), with a third of respondents needing the assistance of a testing partner. Other problems referenced included unstable testing environments (87%) and insufficient time for sufficient testing (85%).
Unsurprisingly, in these circumstances, Applause believes keeping humans in the loop through crowdtesting remains an effective solution to ensure quality, and a third of respondents agree, continuing to do so.
“To meet increasing user expectations while managing AI risks, it’s critical to assess and evaluate the tools, processes and capabilities we’re using for quality assurance on an ongoing basis – before even thinking about testing the apps and websites themselves,” said Rob Mason, Applause’s chief technical officer.
“Agentic AI requires human intervention to avoid quality issues that have the potential to do serious harm, given the speed and scale at which agents operate. The trick is to embed human influence and safeguards early.”
More than 2,100 software testing and development professionals participated in the survey, which is now available online.

It’s not surprising that Applause’s report highlights the need for human oversight when working with AI agents to prevent serious harm.
OpenAI’s co-founder, Greg Brockman, and Codex engineering lead Thibault Sottiaux discussed its importance in Monday’s episode of “The OpenAI Podcast.” In their discussion on what happens when AI becomes a true coding collaborator, as they discussed the new GPT-5 Codex agents, Brockman emphasized the importance of humans "in the driver’s seat,” alongside Codex.
Sottiaux added that their commitment to maintaining a safe environment involves knowing when the agent needs humans to steer or approve actions.
“We're going to be continuing to invest a lot in making the environment safe, invest in understanding when humans need to steer, when humans need to approve certain actions and in giving more and more permissions so that your agent has its own set of permissions that you allow it to use,” he said.
TOGETHER WITH PACASO
Final hours to join big-name investors on this “unlisted” stock
When the former Zillow exec who sold his first company for $120m starts a new venture, people notice. Thousands of people; in fact, 12,000+ regular investors have already joined the early investors behind Uber, eBay, and Venmo as early-stage Pacaso investors.
Disrupting the real estate industry once again, Pacaso’s streamlined platform offers co-ownership of premier properties, revamping the $1.3t vacation home market.
And it works. By handing keys to 2,000+ happy homeowners, Pacaso has already made $110m+ in gross profits in less than 5 years. They even recently reserved the Nasdaq ticker PCSO.
HUMANOID ROBOTS
Figure AI inches closer to mass deploying humanoid robots

Humanoid robot maker Figure AI is one step closer to mass deployment of general-purpose humanoid robots in the real world.
The company, which emerged from stealth in 2023 and is backed by major players including Microsoft and Nvidia, announced this week it has exceeded more than $1 billion in committed capital through its series C funding round, raising the company’s valuation to $39 billion.
Figure said the milestone is critical to unlocking the next stage of growth for humanoid robots, by scaling out the company’s Helix AI system for embodied intelligence and BotQ manufacturing.
In February, Figure’s founder Brett Adcock announced in a post on X that Figure severed ties with OpenAI. This decision came a year after OpenAI invested in Figure’s $675 million series B funding round. Adcock cited that Figure “made a major breakthrough on fully end-to-end robot AI, built entirely in-house.”
Earlier this year, Adcock said he expected his company to ship 100,000 humanoid robots over the next four years. It’s well on its way.
Figure plans to use the latest funding to enhance its development of embodied intelligence by scaling humanoid robots for home and commercial use, building next-generation GPU infrastructure to speed up training and simulation, and launching advanced data collection efforts to improve how robots understand and operate in complex, dynamic environments.
In January of 2024, Figure signed BMW as its first commercial customer, bringing its humanoid robots to the automaker’s factory floor. The following November, Figure started testing its Figure 02 humanoid in BMW’s Spartanburg, South Carolina, plant.
Last April, Figure introduced natural walking to the Figure 02 humanoid robot.
TOGETHER WITH DELL
Dell’s Latest PC Brings AI To Your Desk
Advanced AI may be the future, but you won’t be able to make the most of it if your computer is living in the past. For a new generation of ideas, we need a new generation of PC’s… and that’s exactly what Dell has done with their latest and greatest: The Dell Pro Max with GB10.
The GB10 raises the bar for AI developer PCs thanks to:
A powerful NVIDIA software stack and Grace Blackwell superchip
The ability to support up to 200Bn parameter models
One Petaflop of FP4 computing power
In other words, it’s a pint-sized PC companion that can turn any desk into developer heaven – and it’s available now. Try the Dell Pro Max GB10 for yourself right here.
FUNDING
Invisible raises $100M, now worth $2B

As more companies jump on the AI bandwagon, Invisible Technologies is gaining momentum announcing it raised another $100 million in funding. The investment brings the total raised by the start-up to date to $144 million, with the latest round led by a new company called Vanara Capital, which spun off from private equity firm TPG in August.
Arguably even more noteworthy than the nine-figure sum raised is the valuation of the company itself, reported by Bloomberg to be more than $2 billion, according to unnamed people familiar with the matter.
What attracted the interest of Vanara, as well as other participants, including Greycroft and Acrew Capital, and led to Invisible being named the second-fastest-growing AI company on the 2024 Inc. 5000 list, is its ability to provide the building blocks for AI data training. Or, as Invisible succinctly puts it, “making AI work in the real world.”
It does this via its end-to-end AI Software Platform, which consists of five modular components that enable companies to organize, clean, label and map their data.
Of these components, it is one termed “Expert Marketplace” that has attracted plenty of attention. This gives companies working on Large Language Models access to human experts in specific fields who deliver customized training data and annotations for their AI.
Hayden Lekacz, co-founder of Vanara, said “Invisible is optimizing the adoption of enterprise AI at scale, building the foundational layer of the AI ecosystem that enterprises need, but few have figured out.”
He added that the company “does a much better job at delivering high-complexity, high-difficulty work products to their customers” than rivals.
The funding will be used to further develop the AI Software Platform, with Invisible confident it can add to a client base that includes several of the world’s leading AI model providers, including Cohere, Microsoft and AWS.
LINKS

Clone Robotics’ life-like humanoid robot comes to life.
AI chip startup Groq Inc. raises $750 million, valued at $6.9 billion
Google simplifies sending and receiving money for various AI apps
House Committee asks Discord, Steam and Twitch CEOs to testify at Oct. 8 hearing
Google’s Discover gets upgrade, allowing you to customize your feed
Lyft partners with Waymo to bring robotaxis to Nashville next year

Guidde: AI-powered tool to auto-generate step-by-step video guides, docs, and voiceovers with a free browser extension*
Amazon Seller Assistant: Amazon’s seller assistant received an agent-based AI update to help sellers launch, manage and grow their businesses.
Ask Gemini AI Assistant: Rolling out in Google Meet, Ask Gemini can help summarize discussions, highlight key takeaways, create action items and fill you in on what you missed.
YouTube’s New Creative Tools: YouTube has partnered with Google DeepMind to bring new generative AI tools to its YouTube Shorts creators for easier video clip creation with sound for the first time.

OpenAI: AI Product Counsel
Figure: Robotic Operations Technician
Ford Motor Co.: Data Scientist
SGS Consulting: Artificial Intelligence Engineer
A QUICK POLL BEFORE YOU GO
AI is currently utilized in 60% of software testing. Is human oversight still necessary? |
The Deep View is written by Faris Kojok, Liz Hughes and The Deep View crew. Please reply with any feedback.
Thanks for reading today’s edition of The Deep View! We’ll see you in the next one.

“[The other image] was chosen because of the blurred background, presumably intended for effect.” “It was the out of place yellow flower hanging down, looked like real life!” “The field of focus gave it away. In the fake, the orange flower should have been as blurry or blurrier than the orange pumpkin.” |
![]() | “Just couldn't pick anything obviously generated. The table in the other image looked odd.” “What kind of flowers are those in the other image? I'm no botanist but come on!” “The lighting didn't seem right” |

Take The Deep View with you on the go! We’ve got exclusive, in-depth interviews for you on The Deep View: Conversations podcast every Tuesday morning.

If you want to get in front of an audience of 450,000+ developers, business leaders and tech enthusiasts, get in touch with us here.
*Indicates sponsored content
*Pacaso Disclaimer: This is a paid advertisement for Pacaso’s Regulation A offering. Please read the offering circular at invest.pacaso.com. Reserving the ticker symbol is not a guarantee that the company will go public. Listing on the Nasdaq is subject to approvals.