When James Cameron first released Terminator in 1984, many believed he was a prophet, especially since the film about a killing cyborg engineered by AI was inspired by a nightmare, which he recorded through a painting. The Arnold Schwarzenegger film, which went on to become a classic and spawned multiple sequels, was perceived to be a warning about the dangers of technology – a theme that’s been present in many sci-fi films since. Now, in 2025, that idea could become a real danger as Google begins experimenting with AI-powered robots with Gemini Robotics. Worst of all, it’s powered by Google’s own Gemini – the AI which used the internet, specifically Reddit and Quora, as a learning tool. Is this a recipe for disaster, or have we seen too many sci-fi films?
Google Has Become More Than A Search Engine
In its nearly 30-year existence, Google (now operating under Alphabet Inc.) has provided the world with a plethora of services, devices and more. What started as a simple algorithm-based search engine, which started as BackRub, quickly became a behemoth of a technology company that would seep into your everyday existence.
As with many successful companies, Google’s continued pivot has allowed it to maintain its stronghold as a global force (declared a monopoly by the United States Department of Justice in 2024). Always ready to explore the next, the company has continued to push new technology and services, for better or worse, even going as far as acquiring startups worth billions as a means to this end. Publicly releasing its first AI model in 2023 under its subsidiary DeepMind, Google this week announced the extension of that in the form of Gemini Robotics – an advanced vision-language-action model.
No, Google, A Dog Has Never Played In The NHL
DeepMin released the first public version of its AI model, Gemini 1.0, at the end of 2023. The company expanded beyond a standalone engine into an integrated approach. Much to the enjoyment of the meme community, integrating Gemini into Google’s search engine was somewhat of a joke. Coupled with Google’s vacuous decision to change its search algorithm to prioritise results on public engagement platforms like Reddit, Quora and the like, the answers provided by Gemini to simple questions were nothing short of disastrous and comical. Users quickly picked up on these poor results and began asking questions like “how many cigarettes should a pregnant mother smoke per day?” Among the selection of primary answers was the suggestion that a pack a day would be ideal; it was apparent to see how poorly executed the fledgling service was. Check out some of the worst Google AI answers so far.
We could delve further into the disruption that the new search algorithms caused to businesses and sites across the globe, causing many to close or be forced to cut staff due to lowered views and engagement, but that’s an entirely separate discussion. As a result, and with plenty of pushback from the community, Google was forced to roll back much of its significant algorithm changes, which says a lot. Be that as it may, they would never turn away from its AI-based search engine. Gemini was here to stay, with Google already positioned to upsell ‘advanced AI’ services for your business account.
Is Gemini Robotics Skynet’s Little Brother?
Having calmed the waters over the next 12 months, the company launched Gemini 2.0 in December 2024. As part of its continued investment in AI and broader services that spawn from it, DeepMind launched Gemini Robotics.
In short, Gemini Robotics is an advanced vision-language-action (VLA) model, with its backbone built on Gemini 2.0. The technology aims to bring capacity for multimodal reasoning and understanding environments and situations into the physical world. Its intended aim is to allow robots of any shape and size to perform a wide range of real-world tasks, from manufacturing to daily tasks and eventually your in-home personal assistant.
DeepMind has partnered with Apptronik to realise these goals and to build the next generation of humanoid robots. Their approach is based on three fundamentals – Generality, Interactivity, and Dexterity. When it comes to being general, it means that the technology has to be able to adapt to different situations. Furthermore, they must be interactive – understanding and responding to instructions or changes in their environment on the fly. Lastly, they have to be dexterous, performing tasks that humans can generally do with their hands and fingers.
A History of Failed AI
Google has had its fair share of failed projects and investments over the years. However, if you have over a trillion in market value and billions in the bank, it’s much easier and safer to acquire companies that could be seen as competitors, even if you run them into failure a year or two down the line. That hasn’t deterred the heads at Alphabet from continuing this pursuit with several AI-related projects in recent years.
A decade ago, Google launched what was at the time a very optimistic project in Google Glass. While it was viewed as an augmented reality project, it was also the early days of the global push towards AI, which was nowhere near as advanced as it is today. Less than a year after releasing it into retail, Google Glass was shut down. There was a brief revival in 2017, shortly after which it was again shut down in 2019.
Now, for the third time, there is life in this space for the tech giant, following off the back of moderate success of the Apple Vision Pro. Google launched Android XR in 2024, an AI-based service focusing on AR and VR. Unlike its previous attempts, Google is, at the moment, sticking to what it knows – software – with no physical device announced as of yet.
As with many of its other projects, Google cannot afford to be left behind with the likes of Meta (with Ray-Ban Meta AI Glasses) and Apple to make huge enough strides to leave them behind regarding new AI technologies. These three giants have invested heavily in AI, alongside X (formerly Twitter) and many others, so it was a big shock to the system when DeepSeek was released out of China. These shockwaves were felt even in the stock exchange, showcasing how much movement and reliance on AI technologies the world has become and will continue to make strides towards.
Great Leap in Technology or the Next Big Scam – Are AI Robots Going To Kill Us All?
Many tech enthusiasts have some knowledge about the world of robotics. For many decades, Japan was the forerunner in this space. Even before the introduction of humanoid robots, the country had invested heavily in the technology for decades, especially in manufacturing, with companies like Toyota leading the way. In more recent years, Boston Dynamics has been showcasing its progression. While the company has focused more on military use cases, each iteration has been met with mixed feelings – from optimism in its advancements to fear for the same reasons.
However, the most recent announcement in the world of robotics has been the reveal of Tesla Optimus. As with many previous Tesla showcase events, Elon Musk has been well-known for over-promising and under-delivering, which has bordered on outright lies at times. Even before the recent woes in its stock drop off, the big Tesla event was met with enormous scepticism, so much so that the stock began to fall almost immediately at reopening the next day. This was mainly due to an underwhelming event, primarily due to the impact of its remote control, which Musk promised would be starting production in 2025 and public release in 2026. Given the brand’s history of promising features and products well ahead of its time – looking at the Tesla Semi and Tesla Roadster. The latter was announced in 2017 and said to be an upcoming release for 2020, for which it started taking pre-orders. It has now been eight years since the reveal and five years behind schedule, with Musk still pushing regular false updates, the latest of which being a 2025 production start and 2026 release.
That said, many of the other major companies don’t rely on the whims of their owners to rally the market, whether it be to appease investors or calm public perception. The likes of Google, Meta, and Apple have relied on research and development to showcase their advancements in the market before any big reveals. As such, as shown in the announcement video, DeepMind has already made significant strides towards a fully autonomous robotics with Gemini Robotics, visually demonstrating its progress without the need for CGI-heavy content. As a result, the announcement has been viewed in a much more positive light than its competitor, even without the full-bodied robot like Optimus.
But is this the turning point for AI robotics, and will the fears of sentient AI be realised as we’ve seen in many doomsday movies over the years? In short, no.
As it stands, AI has not progressed much further than regurgitating search results in its most basic form or using trends from previous input data to make a few suggestions within some set parameters in somewhat more advanced scenarios. As it stands, the biggest worry when it comes to hurting humans is limited to erroneous programming, especially when you consider the strength of some of these robot arms, where even just shutting down unexpectedly could break the bones in your hands.
We’re nowhere near the robot uprising or apocalypse many sci-fi films have predicted. And, if the likes of Tesla and its autonomous vehicles are seen as leading examples – a technology that can’t even accurately distinguish one object from the next – we won’t be in a position of panic over robots for many years to come. The biggest threat from AI, however, is the increased accuracy of AI deepfakes.
Tell us, what are your thoughts on Google’s Gemini Robotics and AI? Let us know in the comments.
