I hadn’t seen Emiglio in years — in all probability not for the reason that mid-’90s after I would catch glimpses of him on TV and sigh wistfully. However then one evening a pair months in the past, in a bar in Porto, all of a sudden there he was, within the flesh — or ought to I say, the plastic.
Emiglio, you see, is a robotic. A knee-high butler with a bulbous white head, cartoon grin and pink glowing eyes. I spent my childhood eager for him to convey me enjoyable drinks on his little tray whereas I reclined on the couch, glued to back-to-back Animaniacs episodes. Trying again, my want for Emiglio might nicely have been the rationale I dedicated myself to a profession in writing about know-how.
Learn extra:Â CNET Is Choosing the Best of CES 2026 Awards
Like many fantasies people harbor about robots, this one was ill-thought-out. Who would have made the enjoyable little drinks and balanced them on Emiglio’s tray? How would he have recognized the place to take them? And who, in the end, would have helped him navigate two rooms from the kitchen to me?
This Emiglio was a little bit worse for put on, however then he was dwelling in a bar.
Most of the points that stored Emiglio from being a genuinely helpful butler are the identical limitations real-world robots nonetheless battle with at present. As a lot as I hate to confess it, Emiglio was little greater than a glorified remote-control automobile with a face, needing human help to do virtually something.
The identical is true of Neo, the humanoid house helper robotic that went viral in late October however nonetheless requires teleoperation by a human. The 2 robots are separated by greater than 30 years, but their real-world usefulness and skill to function autonomously really feel equally disappointing.
A query I ask myself annually as I return to CES — the enormous tech commerce present the CNET crew decamps to each January — is when the various robots I’ve met there’ll lastly show worthy of a spot in our properties.
Overcoming the AI bottleneck (and why VLAs matter)
“The primary impediment between us and genuinely helpful house robots is the AI,” famend pc scientist Ben Goertzel mentioned after I sat down with him on the tech-focused Internet Summit in Lisbon final month. The bodily capabilities of robots have progressed dramatically between Emiglio and Neo. What’s holding again Neo, and different house robots, is in the end the smarts.
The AI breakthroughs we have witnessed over the previous few years do pave the way in which for change on this regard. Massive language fashions developed by firms like OpenAI, Google and Anthropic allow us to have extra nuanced conversations with our know-how, which is especially compelling within the case of emotional or companion robots.
It is doable that this 12 months at CES, we might see an organization combine extra superior AI right into a robotic idea it has already showcased, mentioned Ben Wooden, chief analyst at CCS Perception. He instructed Samsung might construct on its rolling projector robotic, Ballie, by working intently with Google — because it already does on telephones — to create a next-generation model outfitted with Gemini, for instance.
“Generative AI is permitting extra pure language interplay with clever gadgets,” mentioned Wooden, however this is identical whether or not you are speaking to a sensible speaker, a robotic vacuum cleaner or a humanoid.
What’s extra helpful to robots than LLMs are developments in vision-language-action fashions, which, because the identify suggests, are a sort of AI that enables for the enter of a mixture of photos and phrases, and the output of actions. For robots navigating a bodily area, this mixture is crucial and can actually set them aside from different AI-equipped gadgets.
“Extra superior fashions for robotics blended with extra succesful and extra deeply built-in generative AI might see some extra clever use circumstances coming to both high-end humanoid science fiction robots or a few of the extra sensible robots,” Wooden mentioned.
There’s been debate over whether or not a breakthrough for robots would require us to unlock AGI — a hypothetical degree of AI superintelligence. Goertzel, who works each on superintelligence and robotics, does not assume that is the case. VLAs are getting so good, we do not want AGI to make first rate house robots, he advised me.
Trying past the humanoid type
The massive problem for house robots is that each house is completely different.Â
Locations equivalent to inns, faculties and hospitals share sufficient similarities that robotic navigation might be kind of standardized. However growing robots for enterprise and business, the place they’re going to be doing repetitive duties in predictable environments, is completely completely different from coaching a robotic that you might put into the wildly diverse layouts of properties.
Nonetheless, some try. The crew at Sunday Robotics, primarily based in California, is coaching its humanoid robotic, Memo, with knowledge offered by households throughout the US that use high-tech gloves to seize the intricate motion of their fingers as they perform family duties. It is an bold method to getting ready robots for household life, and if Sunday Robotics can stick with its desired timeline, it might nicely be one of many first firms to deploy non-teleoperated humanoids in individuals’s properties.
However for some, there’s an actual query mark over whether or not we needs to be aiming for humanoid house helpers in any respect.Â
“If I take into consideration simply on a regular basis stuff, like home tasks, a humanoid physique just isn’t optimum,” Goertzel mentioned. “If I take into consideration in our kitchen at house, my spouse wants me to achieve the excessive issues, and I do not like crawling round within the floor to get the low issues proper, as a result of we’re barely completely different heights. Why construct that downside right into a robotic?”
As a substitute of equipping one costly house humanoid with the identical impediments we battle with as people, he envisions a networked system of smaller, extra sensible bots that may work together and are designed to excel at particular duties.
There are alternatives for established tech firms to leap in right here, whether or not that is Samsung with Ballie and Apple with its mysterious rumored house robotic plans, or firms equivalent to Qualcomm, which might be at CES and will focus on its personal robotics plans on the present.Â
Qualcomm already makes chips for automobiles (that are shut relations of robots, particularly of their autonomous type) and a complete vary of small shopper electronics that maximize energy whereas offering lengthy battery life and AI capabilities. At Internet Summit, CEO Cristiano Amon advised me that he noticed robotics as “an unbelievable alternative.”Â
“We’re excited,” he mentioned. “Whether or not from enterprise to shopper, I believe the kind of silicon that we develop for telephones and for [edge computing] is the right silicon for robots.”
The sensible robotic: Process-specific bots
Many people have already begun to put money into smaller, task-specific bots by shopping for robotic vacuums, mops and lawnmowers — a longtime class that’ll solely growth additional off the again of CES 2026.Â
“There might be an absolute avalanche of robotic vacuum cleaners and mops and robotic garden mowers,” Wooden mentioned. Even so, he famous, you must have “the correct of home” for them to work optimally.Â
Nonetheless, CCS Perception analysis means that 15% of households throughout the US, UK, Spain, France and Germany intend to purchase a robotic vacuum cleaner in 2026. They are not the best or cutest robots, however they do win out on the subject of their usefulness and what persons are truly prepared to spend their cash on.
As for a humanoid house robotic? “Fairly frankly, it is years away,” Wooden mentioned. “Folks love the thought of it, nevertheless it’s a protracted, lengthy, good distance from being one thing that folks would have of their properties and even need.”
The pressing query of privateness and security
His prediction aligns with that of Boston Dynamics CEO Robert Playter, who advised Euronews at Internet Summit that he believes robots will not be in our properties for no less than 5 to 10 years. (This from somebody whose firm makes the acrobatically inclined Atlas humanoid and the intimidating Large Canine, each of which had the US army sniffing round.)
There are a whole lot of causes you won’t truly need that futuristic house robotic, from sensible concerns equivalent to area and usefulness, to greater issues round privateness, security and value (Neo is priced at $20,000, and Sunday Robotics advised me Memo might be a “high-end” product).
In November, robotics researchers at Carnegie Mellon College published a paper declaring that standard AI fashions aren’t but able to energy robots because of points starting from bias and discrimination to unsafe bodily conduct.Â
The examine, which analyzed ChatGPT, Gemini, Copilot and HuggingChat, discovered that the majority fashions have been prepared to approve instructions that may compromise individuals’s mobility aids, permit a robotic to brandish a knife, take nonconsensual pictures or steal bank card info.
“If an AI system is to direct a robotic that interacts with weak individuals, it should be held to requirements no less than as excessive as these for a brand new medical system or pharmaceutical drug,” mentioned co-author Rumaisa Azeem, a analysis assistant within the Civic and Accountable AI Lab at King’s School London. There’s an pressing want, she mentioned, for complete and routine threat assessments earlier than these AI fashions are put into robots.
Only a week after the examine was printed, startup Determine AI was sued by a whistleblower, who warned that the corporate’s humanoid robots might “fracture a human cranium.”
As compelling as they’re to watch, robots equivalent to Elon Musk’s Tesla Optimus and Boston Dynamics’ steady of bots generally is a little unnerving. But when these types of robots are on show at CES this week, the main target will possible be on the talents they’ll convey to industrial settings relatively than to the house.
Robotics might be a “mega pattern” on the present, based on Wooden, nevertheless it stays to be seen whether or not any of the robots on show will rise above robotic vacuums, pool cleaners or cute however one-note companion gadgets to develop into a real family important.
As for me, I nonetheless dream of proudly owning Emiglio — much more so after lastly assembly him. It is tempting to go to eBay to see if anybody is promoting this ’90s robotic toy that my profession is constructed upon, however perhaps it is higher to carry out for the actual factor. For now, I am going to maintain hoping that the actually succesful (and secure) robotic butler of my childhood goals can at some point develop into a actuality.
