Peter Morgan raises some of the questions we will be wrestling with over the next few years as AI makes its way in society.
Artificial Intelligence has crossed a threshold. What was once confined to research labs and speculative fiction now permeates every aspect of life – business, science, creativity, even governance. The pace of progress is unlike anything we have seen before. Generative AI models can write software, produce art, design drugs, and increasingly act as personal and professional collaborators. As we look ahead, the question is no longer whether AI will shape the future – it is how, and in what directions. This article looks at where we are today, the paths that will determine the future state and the factors that may determine whether that future state is utopic or dystopic.
The Current Inflection Point
AI today feels simultaneously astonishing and incomplete. Systems like ChatGPT, Claude, Gemini and LLaMA demonstrate reasoning, creativity, and adaptability once thought impossible for machines. Although greatly improved from just a few years ago, these models still hallucinate, lack robust grounding in the real world, and require enormous computational resources. They are brilliant but fragile, capable but unreliable.
We are at an inflection point, similar to the dawn of the Internet in the 1990s or electricity in the 1880s. AI’s promise is vast, but so are the open questions: how to scale, how to integrate safely, and how to distribute the benefits equitably.
Pathways of Progress
Beyond Scaling
Much of AI’s progress so far has been powered by scaling laws: train bigger models on more data with more compute and they get better. This strategy may be reaching its limits – both economically and physically. The next breakthroughs may not come from sheer size, but from:
- New architectures (for example, neuromorphic chips, hybrid neural-symbolic systems)
- Specialised training regimes (agentic workflows, reinforcement from human feedback, fine-tuning on proprietary data sets)
- Energy-efficient hardware (analogue and/or optical computing)
In short, we are moving from brute force to intelligence in design.
AI as Autonomous Agents
Generative models today are often passive: they wait for prompts. But the future may lie in AI agents that can act, plan, and coordinate across digital and physical environments. Imagine AI scientists reading the latest research, proposing hypotheses, running simulations, and even instructing physical lab robots. Or AI assistants booking travel, negotiating with vendors at work, and monitoring your health in real time.
The shift from models to agents is comparable to the leap from static websites to interactive platforms in the early web. It opens new horizons and new risks, especially around autonomy and control.
The Convergence with Robotics
AI has so far been largely digital. But when paired with robotics, it enters the physical world. Advances in dexterous manipulation, reinforcement learning in simulated environments, and the falling cost of actuators suggest we may soon see useful general-purpose robots.
From elder care to warehouse logistics, the labour market could be transformed. The promise is liberation from drudgery; the fear is mass unemployment. The reality will be a transition from human to digital workers, both white collar and blue. How we navigate this transition will depend on policy, adoption speed, and governance. A whole new economic structure may be needed, including a universal basic income for displaced workers.
Scientific Discovery at Machine Speed
AI’s impact on science may be the most profound. Already, AI systems are designing new proteins, accelerating drug discovery, and predicting material properties. The combination of large-scale data analysis, generative design, and automated experimentation could compress decades of research into years.
This is not about replacing scientists, but about augmenting their creativity. An AI that can scan millions of papers, generate hypotheses, and propose experiments will change how knowledge is created. We may see entire fields – synthetic biology, fusion energy, advanced materials – leap forward at an accelerated pace.
Creativity, Culture, and Meaning
The arts are also in flux. AI can now compose music, write prose, and generate compelling images and videos. Some see this as dilution of human originality, while others see it as an augmentation. In the future, creativity may become more about curation, orchestration, and direction – deciding what to make rather than labouring over execution.
But this raises deeper questions: What is the value of human creativity when machines can flawlessly imitate it? Will audiences crave the authentic, the flawed, the human touch? Or will they embrace infinite abundance of AI generated art? The answers will shape culture probably even more than the invention of photography or cinema once did.
The Economic Shift
Automation and Employment
The automation debate is decades old, but AI raises the stakes. Unlike previous waves of technology, which mainly replaced physical labour, AI can increasingly automate cognitive, white collar work. Legal research and case development, software engineering, financial analysis and risk management, along with greater parts of the medical profession are fields once thought immune to automation but are now vulnerable.
Some roles will vanish, others will transform, and entirely new professions will emerge (AI ethicists, prompt engineers, human-AI interaction designers). The net effect will depend on how quickly economies adapt, how wealth is redistributed, and how humans reimagine their place in the value chain.
The Intelligence Economy
We are moving into an “intelligence economy” where access to machine cognition becomes as fundamental as access to energy or the Internet. Nations, corporations, and individuals with more AI capacity will have disproportionate advantages. This will reshape geopolitics, supply chains, and the balance of power.
Computation—data centres, processors, and energy—will become the new oil. Expect to see intense competition around semiconductor supply, energy efficiency, and cloud platforms. AI won’t just change the economy; it will become the economy.
Ethical and Societal Crossroads
Safety and Alignment
As AI becomes more autonomous, agentic and capable, ensuring its alignment with human goals becomes critical. Misaligned incentives, biased data, or unforeseen emergent behaviours could cause significant harm. Researchers are racing to develop methods for interpretability, alignment, and control, but many of these challenges remain unsolved and are a current work in progress.
Inequality and Access
AI is a force multiplier: those with access to advanced models, data, and compute infrastructure gain enormous advantages. Left unchecked, this could exacerbate inequality both within nations and between them.
- Within societies, highly skilled professionals who know how to leverage AI may see their productivity soar, while others risk displacement.
- Globally, wealthy nations with cutting-edge semiconductors, advanced supply chains and research institutions may race ahead, while developing nations struggle to keep pace.
Equitable access to AI will require deliberate policy choices and international collaboration. Otherwise, AI risks widening the digital divide into a permanent civilizational rift.
Governance and Regulation
Governments are scrambling to regulate AI, but regulation lags behind innovation. Striking the right balance between encouraging innovation and mitigating harm will be one of the defining policy challenges of the 21st century. International coordination will be essential, as AI development knows no borders. The coming decade will test our capacity to match technological brilliance with moral wisdom.
Long-Term Horizons
Toward Artificial General Intelligence (AGI)
One of the most hotly debated questions is when will AI reach human-level general intelligence and above? Optimists argue we are a few breakthroughs away; sceptics believe true generality requires more than scaling neural nets. Regardless, progress is undeniable and AI tools like ChatGPT, Gemini and Claude already equal or surpass humans in many significant areas. If AGI arrives, it could be the most transformative event in history – equal parts opportunity and existential risk.
Human-AI Integration
Equally transformative is the possibility of human-AI symbiosis. Brain-computer interfaces, augmented cognition, and AI companions may blur the line between human and machine. Rather than being replaced, we may become more than human – extended, enhanced, and interconnected.
A Post-Scarcity Future?
In the far horizon, if AI can automate science, engineering, and production, we may enter a post-scarcity era where goods, services, and knowledge are abundant. This raises utopian visions of liberation from labour and dystopian fears of control by a few. The outcome will depend on governance, values, and the choices we make now.
Conclusion: Steering the Future
AI is not destiny, rather it is a tool. The trajectory of AI is shaped not only by technological advances, but by human choices—economic, political, ethical, cultural. We can choose to build systems that amplify creativity, expand access, and solve humanity’s great challenges. Or we can allow AI to exacerbate inequality, concentrate power, and destabilize society.
The future of AI is therefore not just a technical question, it is a civilizational one. What comes next for AI is, ultimately, what comes next for us. The tools we are building will shape not just how we live, but who and what we become.
If the last decade was about the surprise of what AI could do, the next decade will be about deciding what AI should do. That is the true challenge of our time.
Peter Morgan founded the AI consulting company, Deep Learning Partnership, in 2016 to carry out his mission of helping businesses become more innovative and efficient through the use of AI. He works with companies from startups to SMEs to global corporations delivering training and helping them implement AI into their workflows. He is currently Head Tutor on two Oxford University courses on AI for business executives.
This article is also available in the special AI issue of Computers & Law, which is available to download here.
