AI Shifts Everything At Once

Miriam Meckel und Léa Steinacker (photo: Stephanie Pistel)

Once a vision of the future, artificial intelligence is now everywhere and available to everyone. As a foundational technology, AI will enter every aspect of our lives. Its potential is vast but lies in how we will interact with it.

Interview with Miriam Meckel and Léa Steinacker, authors of the book, ‘Everything Everywhere All at Once: How Artificial Intelligence Is Changing Our World and What We Can Gain from It’ (Alles überall auf einmal: Wie Künstliche Intelligenz unsere Welt verändert und was wir dabei gewinnen können).

"Everything everywhere all at once. How artificial intelligence is changing our world and what we can gain from it" — “Everything everywhere all at once. How artificial intelligence is changing our world and what we can gain from it”

Before you read the interview, a few words from the author.

Even though it’s only March, I’ve just read the most exciting book of the year on artificial intelligence. It explains the historical background of AI, what has happened since November 2022, when ChatGPT was released, and what it all means for us, the people.

The authors address universal questions: How will AI enable us to expand our human intelligence? How will AI affect business, human relations, and the geopolitical situation? Can we trust a solution that only simulates thinking but really only predicts the next word in a sentence? The readers receive facts, research-based conclusions, and a thorough analysis of the technology with a connection to psychology and social sciences.

According to the authors, the era of co-creation with AI has begun. Since AI is a superior technology, its expansion will be rapid. We will have to learn how to work and coexist both with its outstanding capabilities and threats. But like any other technology, AI prompts us to ask ourselves questions: What makes us human when machines can be creative?

However, the authors do not delve into dystopian visions. Even if AI transcends the Internet and data from books and encyclopedias to explore the world through robotic bodies and wearable devices, the essence of being human remains unique. We live in symbiosis with millions of bacteria in the gut, which form a second human brain. Checkmate AI!

Enjoy the interview with Miriam Meckel and Léa Steinacker.

Artur: Will a perfectly personalized virtual doctor chatbot be able to gain the same trust and build the same relationship that develops between a patient and a human doctor?

Léa: Likely not the same – but that doesn’t have to make it a less meaningful relationship. As our interactions with AI increasingly rely on voice and gesture over screens and typing, we could be headed towards a scenario as seamless as in the movie.

But we don’t need to look at far-away science fiction to see the evolving relationship between humans and AI. AI chatbots are already being employed in therapy, with mixed results so far. In one cross-sectional study, published in JAMA Internal Medicine, of 195 randomly drawn patient questions from a social media forum, a team of licensed healthcare professionals compared physicians’ and chatbots’ responses. The chatbot responses were preferred over physician responses and rated significantly higher for both quality and empathy. However, in another experiment, an online emotional support service found that people rated a chatbot’s messages as more helpful than those written by humans, but only as long as they did not know a chatbot was involved.

Similarly, the quality of diagnosis has sometimes been proven to be better through AI, while in other instances, dangerous information has been produced that could potentially be harmful to patients. Evidently, the human physician in the loop remains essential.

Miriam: Talking about doctor chatbots only for conversing with patients might be missing the point, actually. There’s a lot more potential in using AI to free up physicians’ time so they can focus more on patients, not less. Here, the idea isn’t to replace doctors and therapists with AI but to take over the repetitive bureaucratic tasks that eat up their time. For instance, AI can help schedule doctor appointments and sort mail and other documents. It can also support doctors in their medical tasks, as with image recognition AI, for example. Our approach to AI should always be to amplify our human strengths and connections, not weaken them.

Artur: The example of a disruptive technology like smartphones shows that we have overestimated the long-term risks of the technology and underestimated the benefits. Will the same be the case with AI?

Miriam: We often exaggerate the immediate impacts of new technologies while overlooking their long-term consequences—a point futurist Roy Amara highlighted decades ago. With AI, there are indeed legitimate concerns regarding privacy, autonomy, and job displacement, or, rather, the displacement of tasks, that we need to address. But these concerns should not overshadow the potential benefits. AI has the power to significantly improve our ability to diagnose diseases earlier, enhance educational outcomes, and make our cities smarter and more sustainable. The key is to strike a balance, using AI to augment our capabilities while remaining mindful of its ethical and societal implications.

Léa: There are risks, certainly, some of which are difficult to quantify. Take the issue of deepfakes, for example, which can distort our perception of reality and even influence elections. In India, deepfakes have started to become a common tactic for undermining opponents, and we’ve seen instances, like the fabricated arrest of Trump, that never happened. However, the manipulation of images is not a new phenomenon; it even predates the digital era. This is why some advocate for the importance of limiting the spread of such images by catching them early and holding social networks accountable for their role. Major tech platforms, including TikTok, X, and Facebook, are poised to play a crucial role in identifying AI-generated content to safeguard the upcoming European election from disinformation.

We definitely need safeguards to explore AI safely—and education for ourselves and others is essential. Yet, we should not be swayed by those who portray AI as an imminent threat to human survival. Often, such arguments serve to divert attention from the practical challenges of AI that we can—and should—address.

Artur: The book “Alles überall auf einmal” has a subtitle: How artificial intelligence is changing our world and what we can gain from it. What arguments support such an optimistic outlook?

Miriam: The inspiration behind our book’s title comes from the outstanding film “Everything Everywhere All at Once.” In the movie, Evelyn Wang, a middle-aged Chinese immigrant running a laundromat with her husband, discovers the existence of multiple parallel universes. The idea is that every decision we make creates a new alternate universe. In a way, AI offers us numerous opportunities to live and work in ways we might not have envisioned growing up. You can look at AI as a challenge to the status quo, or you can see it as opening up all these different possibilities, these different universes that you could step into. There’s clear evidence that AI can significantly enhance our productivity and efficiency when applied correctly. Take, for example, a study that found consultants using AI to be substantially more productive—they managed to complete 12% more tasks on average and did so 25% faster, with the quality of their work being over 40% higher compared to a control group.

Léa: And to bring it back to the film, there is a scene with a couple of speaking stones. Those were AI-generated, and the film’s visual effects artist, Evan Halleck, said that it took him much less time to create that scene than it would have without AI. With tools like Runway ML or Sora, OpenAI’s video generator, creating film scenes can now be a matter of minutes or even seconds for anyone. This is incredibly impressive and holds a lot of promise for the future of our work lives. However, the challenge lies in using the time AI tools free up in a meaningful way. Google’s Duet AI, for instance, made people more productive to the point where they were simply writing 30-40% more emails, just more quickly. A case like that makes you wonder if we’re actually increasing productivity or just inflating our workload. Perhaps, like Evelyn Wang, we need to decide sooner rather than later which universe we really want to live in – and use AI to help us get there.

Artur: Are regulations still able to prevent the harmful effects of AI? Are we in danger of a scenario similar to climate change? Example: The Paris Agreement on climate change was created too late and, as we can see, is adhered to differently worldwide.

Léa: I would actually argue that if there’s any lesson to be learned from the Paris Agreement, it is that it’s never too late to take action and that we need to seek international alignment if we wish to regulate a common good effectively.

At the same time, it’s important to lead by example locally, too. With the AI Act, the EU has made the first comprehensive multinational attempt at regulating AI. While imperfect, the AI Act at least sets an example of how we might regulate AI in practice. Other countries are no doubt watching the EU’s approach closely. That said, just like there’s no one way to protect the environment, there’s no one way to protect fundamental rights. The U.S., for instance, is pursuing a sectoral approach instead. While this has been criticized in the realm of data protection, it may be a more effective way of regulating AI. This is because the types of AI systems deployed and associated harms can vary dramatically from one sector to the next. Unlike the EU, the U.S. is also investing much more heavily in AI innovation, which is an important complementary measure to the protection of rights.

Miriam: The EU AI Act could actually be seen as a testament to the opportunity of different states, each with their own interests, to come together to agree on a set of rules around AI. Yes, these regulations are imperfect, but that’s the nature of compromise when multiple parties are involved. It is definitely not too late to address the harmful effects of AI. But our strategy shouldn’t be limited to just AI regulation. The responsibility also lies with us to build the necessary skills that enable responsible AI use. As citizens and consumers, we need to enhance our AI literacy to understand the challenges AI might pose. This includes knowing when and how to verify facts, how to question AI output, and appreciate what constitutes a well-written text. Editing AI-generated texts well, like those from ChatGPT, doesn’t require any less skill than editing human-written texts; if anything, it demands more. Moreover, protecting our elections might mean shifting our regulatory efforts back to social media, as we’ve seen with the Digital Services Act, rather than solely focusing on AI systems. Generally, the harms AI can produce are not exclusive to AI but may also arise in the context of other complex software systems.

Artur: What kind of healthcare would you like to see in an era where AI is “everywhere all at once”? What should stay the same, and what might change?

Miriam: My vision is for a healthcare system where accessibility is not just a goal but a reality. AI could help us erase the boundaries that currently limit healthcare access. This could mean deploying AI-driven solutions to bring medical advice and preventive care to the most remote corners of the world, ensuring that distance or socioeconomic status no longer determines access to quality care. It could also mean scheduling doctors’ appointments in a way that everyone gets a chance to make an appointment and better distribute doctors’ limited time. AI could offer everyone, everywhere, the chance to benefit from advanced healthcare services. But for this vision to come true, it’s critical that we invest in the necessary infrastructure and digital literacy, making sure that these technological advances are universally accessible and not just another layer of advantage for the already privileged.

Léa: I believe personalized care should be at the heart of healthcare transformation. With AI, we have the unprecedented opportunity to tailor healthcare to each individual’s unique needs, from genetic predispositions to personal health histories. This approach could dramatically improve how we manage and treat chronic diseases, allowing for earlier interventions based on predictive analytics and offering treatment plans that are customized for the best possible outcomes. If we manage to safeguard patient privacy and ensure that the data used to personalize care is protected with the highest standards of security, then this personalized approach promises better health outcomes and could help us lead longer and healthier lives.

Artur: With a vision of more accessible and affordable healthcare, personalized medicine, and healthier lives, I wholeheartedly agree that we should embrace the use of AI. Thank you for your time!