Schmidt Warns AGI May Arrive Before 2028

Former Google CEO Eric Schmidt is sounding the alarm: artificial general intelligence — the kind of AI that can match or exceed human-level reasoning across virtually any domain — could arrive before 2028. The warning, coming from one of Silicon Valley's most influential figures, has intensified an already heated debate about the pace of AI development and whether the world is remotely prepared for what comes next.

Schmidt's timeline is far more aggressive than many mainstream predictions from just 2 years ago. Where researchers once placed AGI decades away, a growing chorus of tech leaders now believes the breakthrough could happen within the current decade — or sooner.

Key Takeaways at a Glance

• Eric Schmidt believes AGI could emerge before 2028, significantly earlier than many prior forecasts
• The prediction aligns with accelerating capabilities seen in models like GPT-4o, Claude 3.5 Sonnet, and Gemini Ultra
• Schmidt has urged governments to treat AI safety and regulation with wartime urgency
• Current AI investment exceeds $100 billion annually across major tech companies
• The gap between 'narrow AI' and AGI is shrinking faster than most experts anticipated
• Safety infrastructure and governance frameworks remain woefully underdeveloped for an AGI-level event

Why Schmidt's Warning Carries Unusual Weight

Eric Schmidt is not a casual commentator. He led Google from 2001 to 2011, overseeing its transformation into one of the world's most powerful technology companies. He later chaired the National Security Commission on Artificial Intelligence, advising the U.S. government on AI strategy and defense applications.

His credibility on this topic is reinforced by his proximity to cutting-edge research. Schmidt has maintained deep ties to AI labs, invested heavily in AI startups, and engaged directly with researchers at DeepMind, OpenAI, and Anthropic. When he speaks about timelines, he does so with insider knowledge that few public figures possess.

What makes this particular warning notable is Schmidt's evolution. Just a few years ago, he expressed more measured views about AGI timelines. His shift toward a pre-2028 prediction reflects what he describes as an 'exponential curve' in AI capability that has surprised even those building the systems.

The Evidence Supporting an Accelerated Timeline

Schmidt's prediction does not exist in a vacuum. Multiple data points suggest the pace of AI advancement is accelerating beyond prior expectations.

Scaling laws continue to hold. Each generation of large language models demonstrates meaningful leaps in reasoning, planning, and multimodal understanding. GPT-4, released in March 2023, showed performance that experts had not expected until 2025 or later. Claude 3.5 Sonnet from Anthropic demonstrated coding and analytical abilities that rival junior professionals in certain domains.

Key indicators supporting the accelerated timeline include:

• Compute investment is doubling roughly every 6-9 months, with companies like Microsoft committing over $13 billion to OpenAI alone
• Synthetic data and improved training techniques are reducing the dependency on human-curated datasets
• Agentic AI systems — models that can plan, execute multi-step tasks, and use tools autonomously — are progressing rapidly
• Multimodal capabilities now allow models to process text, images, audio, and video in unified architectures
• Reasoning benchmarks that once stumped AI systems are being solved at rates that compress previous 5-year forecasts into 12-18 months

The trajectory is undeniable. Whether it leads to true AGI by 2028 remains debatable, but dismissing the possibility outright now seems intellectually irresponsible.

Defining AGI: The Goalpost Problem

One of the most contentious aspects of any AGI prediction is the definition itself. There is no universally agreed-upon benchmark for when an AI system qualifies as 'generally intelligent.'

OpenAI defines AGI as 'highly autonomous systems that outperform humans at most economically valuable work.' DeepMind has proposed a more granular framework with 5 levels of AGI, ranging from 'emerging' to 'superhuman.' By DeepMind's classification, current frontier models like Gemini Ultra and GPT-4o already qualify as Level 1 — 'emerging AGI.'

This definitional ambiguity matters enormously. If AGI is defined as a system that passes every conceivable human cognitive test, we may be further away. But if AGI means a system that can autonomously perform the vast majority of knowledge work — writing code, conducting research, managing projects, analyzing data — then Schmidt's timeline starts to look conservative.

The practical reality is that the economic and societal impact of AI does not require a clean 'AGI moment.' Incremental advances in reasoning and autonomy are already transforming industries. The line between narrow AI and AGI may blur rather than snap.

The Safety Gap Is Growing, Not Shrinking

Schmidt has consistently paired his capability predictions with stark warnings about safety. His core argument is that the institutions responsible for managing AGI-level risk are moving at a fraction of the speed of the technology itself.

Government regulation remains fragmented and slow. The EU AI Act, the most comprehensive regulatory framework to date, focuses primarily on current-generation AI risks like bias and transparency — not the existential challenges posed by AGI. In the United States, executive orders and voluntary industry commitments represent the extent of federal action.

Meanwhile, the AI safety research community, while growing, remains vastly outnumbered by capabilities researchers. Anthropic has arguably made the strongest institutional commitment to safety, dedicating significant resources to its Constitutional AI approach and interpretability research. But even Anthropic's CEO Dario Amodei has acknowledged that the field does not yet have reliable methods for aligning a superintelligent system.

Schmidt has called for several urgent measures:

• Establishing international AI governance bodies with enforcement power
• Mandating safety testing and red-teaming for frontier models before deployment
• Creating 'kill switch' protocols for systems that demonstrate unexpected autonomous behavior
• Dramatically increasing public funding for AI alignment research
• Building monitoring infrastructure to track compute usage and model capabilities globally

The gap between where safety infrastructure stands today and where it needs to be for an AGI-capable world is, by Schmidt's assessment, dangerously wide.

How Other Tech Leaders View the Timeline

Schmidt is not alone in his aggressive timeline, but he is not universally supported either. The AI community remains divided on when — and whether — AGI will arrive.

Sam Altman, CEO of OpenAI, has suggested AGI could arrive 'sooner than most people think,' with some interpretations placing his estimate around 2025-2027. Jensen Huang, CEO of Nvidia, has predicted AGI within 5 years, depending on how it is defined. Demis Hassabis of Google DeepMind has offered a slightly more cautious view, suggesting AGI-level systems could emerge by 2030.

On the skeptical side, prominent researchers like Yann LeCun, Meta's chief AI scientist, argue that current LLM architectures are fundamentally insufficient for AGI. LeCun contends that true general intelligence requires entirely new approaches to world modeling and reasoning that have not yet been invented.

This disagreement is not merely academic. It shapes investment decisions, policy priorities, and public perception. If Schmidt and Altman are right, the world has less than 3 years to prepare for a transformative event. If LeCun is right, there is more time — but potentially less urgency driving safety work.

What This Means for Businesses and Developers

Regardless of whether AGI arrives in 2028, 2030, or 2035, the practical implications of Schmidt's warning are immediate. Businesses and developers should be operating under the assumption that AI capabilities will continue to advance rapidly.

For enterprise leaders, this means accelerating AI integration strategies now rather than waiting for a 'perfect' model. Companies that build robust AI infrastructure today will be best positioned to leverage increasingly powerful systems as they emerge. The competitive advantage will belong to organizations with mature data pipelines, skilled AI teams, and clear governance frameworks.

For developers, the message is equally clear: invest in understanding agentic AI systems, multi-model architectures, and AI safety principles. The developers who can build reliable, safe, and scalable AI applications will be in extraordinary demand as systems grow more capable.

For policymakers, Schmidt's warning should serve as a catalyst. The window for establishing meaningful AI governance is narrowing. Waiting for AGI to arrive before creating regulatory frameworks would be like building fire codes after the city has already burned.

Looking Ahead: The 2025-2028 Window

The next 3 years represent what may be the most consequential period in the history of artificial intelligence. Whether or not AGI arrives by 2028, the capabilities developed during this window will reshape economies, labor markets, national security, and scientific research.

Schmidt's prediction forces a critical question: are we building the future we want, or merely the future that arrives fastest? The answer depends on choices being made right now — in boardrooms, research labs, and legislative chambers around the world.

The former Google CEO's warning is not a prophecy. It is a call to action. And the clock, by his estimation, is ticking faster than almost anyone expected.