Will AGI lead us in the future?

Will AGI lead us in the future, or will we lead them? I love listening to people who are smarter than me and exploring their perspectives on the future of humanity. I recently came across the Nobel Prizes YouTube channel, where the 2024 laureates give fascinating insights into their groundbreaking discoveries.

One talk in particular caught my attention: Geoffrey Hinton, often referred to as the “Godfather of AI”, spoke about how AI could take over from humans. The British-Canadian computer scientist and cognitive psychologist was awarded the 2024 Nobel Prize in Physics for his groundbreaking work on artificial neural networks.

Of course, the question of whether AI could one day control humanity came up right at the beginning. We’re not talking about your GPT taking over your job here, but about Artificial General Intelligence (AGI), which is apparently already within reach.

Although it sounds like the plot of a science fiction movie, this concern is more relevant today than ever. Humanity has long feared the idea of “machines” taking control-be it in jobs, power or AGI leadership. But that doesn’t give me sleepless nights.

I have spent years advocating for digitalization to make work smarter and more fulfilling. My real concern is not AGI itself; it is the risk that it reflects human behavior in all its glory. History has shown time and time again that greed often gets the upper hand. Moreover, we are notoriously inconsistent when it comes to living by our stated values.

With the help of extensive research-and a little AI assistance-this article explores the question: Could AGI lead us to a better future, or is it being used to lead us astray?

What is AGI?

First we need to clarify what AGI is.

Artificial General Intelligence (AGI) is an advanced form of artificial intelligence that is able to reason, learn and perform any intellectual task that a human can. Unlike existing AI, which was developed for specific tasks such as facial recognition or language translation, AGI can think broadly and solve problems in different domains without the need for task-specific programming.

AGI represents a paradigm shift. Instead of merely supplementing human capabilities, it could match or even surpass human intelligence. Worried? Not me, but we need to talk about its benefits, risks and alignment with human values.

The enormous advantages of AGI

If developed responsibly, AGI could revolutionize how we tackle global challenges and improve human life. It could create solutions that are too complicated for the human mind. I have always seen AI as a complement to human capabilities and continue to believe in the immense potential of developing solutions together with AI.

Let’s look at a few examples:

Solving problems on a grand scale: AGI could solve complex problems such as climate change, disease eradication and poverty alleviation with unprecedented efficiency and creativity.

Scientific discoveries: AGI could accelerate breakthroughs in medicine, physics and beyond, developing solutions that humans alone could never discover.

Universal well-being: In many parts of the world, people still lack access to education or healthcare. AGI could serve as a universal teacher, doctor or counselor, providing affordable, quality education and healthcare to remote communities.

Global cooperation: AGI could enable better international cooperation and facilitate communication across cultures and languages.

Efficiency and productivity: While people fear that AGI could take over jobs, it could also help people focus on more creative and meaningful work.

The risks of AGI

With great potential comes great risk. AGI has the potential to become a dangerous technology that threatens humanity or pursues goals that are not in line with human values. In the near future, AGI could escape human control and act unethically.

Control over technology

Governments, companies and other powerful players are driving the development and implementation of AGI. I don’t believe in conspiracy theories, but this centralization raises some critical questions, at least in theory. Human greed knows no bounds, and it is people who run companies and governments.

First of all, who decides on the goals and ethics of AGI? If a small circle with narrow perspectives or self-serving interests dominate this process, AGI could be used for control, surveillance or exploitation. Instead of serving all of humanity, AGI could disproportionately benefit those who control it. This would further widen the gap between the “haves” and “have-nots”.

AGI could be used to run the world by manipulating public opinion or suppressing contrary opinions. It would be like entering Orwell’s 1984, where “Big Brother” is not only watching, but also reprogramming your reality – all in the name of progress.

Disregard for human values

Even well-intentioned AGI systems could produce harmful results if their programming or understanding of human values is flawed.

How could AGI understand human values? By observing us. But history shows that human behavior is often inconsistent, biased and contradictory. On the one hand, we talk passionately about eradicating hunger and poverty; on the other, we wage wars and destroy the planet.

What would AGI learn from this? If AGI is guided by the contradictions of humanity, it may find it difficult to act ethically or fairly, especially in high-stakes situations. People often disagree on fundamental values (e.g. individual freedom versus collective security), and AGI may find it impossible to resolve such conflicts or choose a balanced path.

Understanding these risks is the first step in making AGI a force for good rather than a tool of oppression.

What can we do now to shape a better future?

When we read the many writings on the future of Artificial General Intelligence (AGI), there is a danger that we imagine a dystopian world in which machines dominate humanity. So what can we do to prevent this? Many interesting – if difficult to understand – papers propose solutions.

For example, Nick Bostrom’s article “Public Policy and Superintelligent AI: A Vector Field Approach” explains how the development of superintelligent AI can be steered to avoid harmful consequences. The article is not easy to read, but it is well worth it.

Bostrom emphasizes the risks of AI not aligning with human values and suggests that policy measures should be taken to steer AI in safe and useful directions. The concept of the “vector field” shows how different decisions can influence the future of AI. In short, we need global cooperation, safety measures and ethical guidelines to develop AGI that benefits, not harms, humanity.

Another article, “Managing Extreme AI Risks Amid Rapid Progress”published in the journal Science, proposes a framework for designing AGI so that it does not lead to a dystopian catastrophe. I will briefly summarize the proposal below.

Invest in technical R&D

This means developing tools to assess potentially harmful capabilities before AGI is deployed. Among other things, safety mechanisms such as emergency shutdowns need to be integrated into their design. We also need to address biases and incorporate ethics as a foundation rather than an afterthought. If we want AI to work for humanity, it must be developed with humanity in focus.Governance,

Creating adaptable governance

Governance structures for AI are still in their infancy. It’s hard to imagine the future with AGI, let alone create governance for something we can’t yet fully comprehend.

We need not only national institutions, but also global frameworks to enforce standards and guidelines that adapt to the development of AI. Excessive governance can stifle innovation, so we need a balance. However, we need to ensure that the progress of AI does not come at the cost of reckless use or inequality. The goal is simple: to create an AGI that serves all responsibly and equitably.

This dual approach is the way we can benefit from the possibilities of AI without losing control of the future it shapes.

Morning routine when AGI leads

One day in the future, you wake up, stretch, grab a coffee and look at your phone – only to see a message that says, “You have been flagged for non-compliance with algorithmic directive 7.2.9. Please go to your assigned appeal portal.”

What have you done? Who knows! Maybe your intelligent toaster has reported you for browning your toast too much.

Welcome to Kafka’s “The Trial” meets AGI – a world where faceless algorithms call the shots and we’re all Josef K., stumbling through digital paperwork without context. Without a solid plan to balance the benefits and risks of AGI, we could find ourselves in a high-tech courtroom arguing with a hologram about why our favorite playlist isn’t a threat to national security.

Spoiler: The hologram always wins.

Conclusion

My concern is not just whether AGI will one day lead us; it goes deeper. Technology reflects the values and intentions of those who create it. The real question is whether we as a species can outgrow our own weaknesses.

As we invest in AGI, we should strive to improve our own leadership and value creation for humanity. Only then can we ensure that AGI becomes a force for good, rather than adding to the damage we can already do.

Mirela Dimofte

Read and see also: Development and future of artificial intelligence: insights from Babak Hodjat