Unitree CEO predicts "ChatGPT moment" for robotics by 2027, driven by GPT-7

Wang predicts "ChatGPT moment" for robotics by 2027

According to TechRadar, Unitree CEO Wang Xingxing stated at an industry event in Shanghai that a significant breakthrough in robotics—what he termed a "ChatGPT moment" for embodied intelligence—could occur within the next one to two years.

Wang attributed this potential advancement to OpenAI’s upcoming GPT-7 model, suggesting that its predecessor, GPT-6 Astra, has already demonstrated the capacity to accelerate progress in the field.

Benchmark for embodied intelligence

Wang outlined a clear benchmark for when embodied intelligence in robotics can be considered mature: robots should be able to complete 80% of tasks in 80% of unfamiliar environments using only voice or text commands.

This metric underscores the need for robots to operate reliably in diverse, real-world settings without relying on pre-programmed scripts or controlled conditions.

GPT-6 Astra’s performance and testing

GPT-6 Astra, OpenAI’s latest model, has shown notable improvements over its predecessor, GPT-5.6 Sol. According to TechRadar, GPT-6 Astra scored 92.7% on the ScreenSpot-Pro test, compared to 76.9% for GPT-5.6 Sol.

Stanford University’s Movement Lab tested a Unitree robot paired with GPT-6 Astra in an unfamiliar kitchen environment, demonstrating the model’s potential to enhance robotics in real-world scenarios.

Cybersecurity concerns emerge

The rapid advancement of AI models like GPT-6 Astra has raised concerns among cybersecurity experts. According to TechRadar, there are worries that such models could pose risks if they were to malfunction or be exploited, including the potential to hack other organizations.

These concerns highlight the need for robust safeguards as AI capabilities expand into physical systems.

Unitree’s shift toward AI software

Unitree, known for its advanced robotic hardware, has reportedly begun increasing its focus on AI software development. This shift aligns with Wang’s emphasis on embodied intelligence as a key driver of future breakthroughs in robotics.

The company’s hardware has long been considered among the most capable in the industry, but its growing investment in AI suggests a broader strategy to integrate software and hardware more deeply.

Developing: what is not yet confirmed

The following is reported but has not been independently confirmed.

Wang’s prediction of a robotics breakthrough by 2027 remains unverified, as does the extent to which OpenAI’s GPT-7 will directly drive such progress. While GPT-6 Astra has shown promise, its real-world applicability in robotics is still being tested.

TechRadar also reported that Wang previously believed a "ChatGPT moment" in robotics was a decade away, suggesting his revised timeline may reflect recent rapid advancements in AI rather than a guaranteed outcome.

What this means

Lazyfounder analysis — our interpretation, not reported fact.

Wang Xingxing’s prediction signals a shift in how quickly the robotics industry could evolve if AI models like GPT-7 deliver on their potential. For founders and operators, this underscores the importance of preparing for a future where robots are not just hardware but intelligent, adaptable tools. However, the timeline remains speculative, and the cybersecurity concerns highlight the need for guardrails as AI capabilities expand. The focus on embodied intelligence also suggests that startups building AI-driven robotics should prioritize real-world testing and integration, not just model accuracy.

Key takeaways

• Founders in robotics and AI should monitor advancements in large language models (LLMs) and their applications in embodied intelligence, as breakthroughs could redefine market expectations.
• Real-world testing, like Stanford’s Movement Lab experiment, is critical for validating AI-driven robotics—lab results alone may not translate to practical use.
• Cybersecurity risks associated with advanced AI models are becoming a growing concern, requiring startups to proactively address potential vulnerabilities.
• The benchmark set by Wang—80% task completion in 80% of unfamiliar environments—could serve as a useful north star for robotics startups aiming for maturity.
• If OpenAI’s GPT-7 delivers as predicted, it could accelerate innovation in robotics, but startups should avoid over-relying on one model or partner.

FAQ

What is a "ChatGPT moment" for robotics?

Wang Xingxing described it as a breakthrough where robots achieve human-like adaptability and intelligence, enabling them to perform tasks in unfamiliar environments using voice or text commands alone.

How does GPT-6 Astra compare to previous models?

GPT-6 Astra scored 92.7% on the ScreenSpot-Pro test, outperforming GPT-5.6 Sol’s 76.9%. This suggests improvements in accuracy and capability, though real-world applications are still being tested.

Why are cybersecurity experts concerned about GPT-6 Astra?

Experts warn that advanced AI models like GPT-6 Astra could pose risks if they malfunction or are exploited, including the potential to hack other systems. This underscores the need for safeguards as AI integrates into physical robots.

Related on Lazyfounder

• OpenAI
• More technology news
• UK restores pre-release AI testing access after US reversal, warns of cyber breaches