Prof. Dr. Zander on the Transformative Impact of Passive BCIs and AI on Industries

Prof. Thorsten Zander (TZ): Coming from mainly a research background, I co-founded Zander Labs with my colleague, Marc Grootjen, aiming to redefine the brain-computer interfaces (BCIs) field. A brain-computer interface (BCI) is a technology that processes information from the human brain without relying on muscular activity. While we typically use muscles to communicate information through speech or gestures, a BCI can detect specific patterns in brain activity associated with emotions, states of mind, and interpretations.

Until 2008, BCIs primarily assisted those with severe disabilities to help them communicate, but I envisioned broader applications, in particular in using this technology for healthy people. My entry into the field introduced the concept of passive BCIs, assessing subconscious information without relying on muscular activity. With a BCI, you could observe my current emotional states, like happiness or stress. If I respond with agreement or disagreement to something you say, it reflects the information my mind is naturally producing. This is part of my normal thinking process, assessable through what I term a passive BCI. This concept had a significant impact, altering the main definition of BCIs to now include passive BCIs. This shift sparked significant debates within the BCI research community, leading to the inclusion of passive BCIs in the definition.

Frustrated by the limitations of academic research, we founded Zander Labs to bridge the gap between research and market implementation. Our goal was not to build a large company initially but to showcase the potential of the technology through small projects. The unexpected success prompted collaborations with prominent companies like Volkswagen, Siemens, and others got interested, leading to long-term continuous studies.

Zander Labs evolved beyond our initial expectations, with a focus on bringing technology to the world rather than creating a massive company. As a result, we successfully secured a project with a contract of over $30 million in 2023, a singular entity receiving support for our endeavors. This distinction is crucial, as it involves subcontracting rather than forming a consortium.

We currently have 20 employees and have secured patents in both Europe and the U.S., covering our intended innovations. Through collaborative efforts, we have gained recognition from the German Government, specifically the Federal Agency for Innovation in Cyber Security which funded the 30 million contract mentioned above.

In summary, Zander Labs was founded to advance BCIs beyond their traditional applications, leading to unexpected success and collaborations with major companies in the industry.

(TZ): Certainly! For one of our clients, we collaborate on projects aimed at enhancing both training sessions and daily work for individuals such as pilots. By assessing their brain activity and interactions with technology, we optimize processes to minimize errors and improve overall work quality.

For instance, during training, our system identifies challenging tasks for individuals and tailors the session accordingly. It recognizes when a task is difficult and repeats or provides additional information. In a language-learning scenario, the system gauges immediate comprehension of word pairs and adapts the learning process based on individual understanding.

In operational settings, like a pilot interacting with autonomous systems, our passive BCI detects instances of confusion or miscommunication. This allows the machine to adjust its actions, ensuring better mutual understanding between the technology and the user. A big statement that it may be, the machine can learn to understand the pilot empathically.

This approach lays the groundwork for empathic machines, a crucial step towards addressing nuanced needs in various situations. In a broader context, such technology has potential applications in elderly care, where passive BCIs can make robots neuroadaptive to better understand and respond to individuals' changing moods and needs.

I'm confident in the capabilities of technology for elderly care, but current robots lack empathy. In the last phase of life, human emotions vary, and a robot's understanding is limited. Passive BCIs can make robots neuroadaptive, potentially improving their comprehension of human moods beyond what humans might achieve.

Our work with customers represents an initial stride towards achieving this level of empathic understanding, showcasing the adaptability and potential of our technology in particular automation surprise.

(TZ): Currently, we've demonstrated the effectiveness of our AI-informed technology in laboratory settings, ensuring it's more than just a conceptual idea. The challenge lies in transitioning this technology to real-world applications, emphasizing human factors, engineering, and practicality.

The current process involves time-consuming steps. For example, we're using standard EEG caps with 64 electrodes. Currently, applying gel to each electrode on the scalp takes half an hour with two experts. This bottleneck needs improvement. To address this, we're working on developing user-friendly and fashionable sensors that can be self-applied, as well as universal classifiers for quicker and more convenient BCI calibration.

Privacy and cyber security concerns are paramount. To address this, our project includes a chip for on-device processing, ensuring sensitive brain data doesn't leave the chip. Users have full transparency and control over shared information, enhancing cybersecurity and privacy.

Once these enablers are ready in the next two years, we anticipate a surge in deployable applications, breaking down previous barriers. As for empathy, the timeline depends on the complexity of the desired application. A more empathic robot for elderly care, for example, involves additional steps and might take several more years. However, within three years, we foresee the market introduction of products with enhanced empathic capabilities compared to current technologies, with the full realization of deeply understanding robots requiring further advancements in other domains.

(TZ): Another significant challenge, not previously mentioned, involves context awareness in BCI applications. While detecting emotions like surprise is valuable, it becomes more meaningful when the system understands the context of that emotion.

For instance, knowing that someone is surprised is interesting, but it becomes crucial when the system can deduce the reason behind the surprise. This requires the BCI system to understand the surrounding situation, such as recognizing a bird flying outside the window and knowing where the person is directing their attention.

To address this, we are actively working on a comprehensive hardware and software toolbox to facilitate context awareness. This development environment will enable AI systems to understand the ongoing context, whether in a car or an office, marking a crucial step toward more nuanced and insightful BCI applications.

While there are numerous potential threats and challenges in the BCI and NeuroTech industry, it's crucial to focus on general problem prevention rather than specific concerns. Rather than alarming people with hypothetical scenarios, the emphasis should be on effectively communicating the technology's potential and its positive and negative aspects. When the first trains were introduced, some believed passengers would instantly perish due to the high-speed pressure, causing internal organ failure. There were also concerns about the mind being unable to process information at such speeds, potentially leading to madness. However, those fears didn't materialize. Instead, new issues arose. Transportation systems inevitably come with their own set of risks and challenges.

The responsibility lies with researchers, entrepreneurs, and the entire research community to inform potential users and the general population to make responsible decisions. Drawing a parallel to past innovations like the internet, where early concerns like filter bubbles and fake news were overlooked, the lesson is to avoid repeating the same mistake. The key is to ensure people are aware and capable of making informed decisions.

Having robust laws and regulations is essential, but they won't completely safeguard against users willingly providing sensitive information. A potential concern could be the misuse of BCIs by social networks to gather data on users' reactions to political content, enabling highly targeted manipulation. The exponential empowerment of micro-targeting could lead to a form of mind control influencing political views.

To mitigate such risks, the focus should be on making users aware of these possibilities, encouraging them to make informed decisions about when to use the technology. For instance, users could turn off the system when engaging with political news.

The goal is not to create fear but to improve lives, enhance security, and increase productivity. By acknowledging potential threats and keeping people informed, we can navigate the evolving landscape of NeuroTech. We understand that the real threats might emerge in unforeseen ways. The key is to be prepared, stay informed, and work towards making the future as positive and beneficial as possible.

(TZ): In simple terms, currently, AI functions as a child-like learner who can recognize patterns in the structured world but lacks true understanding. Drawing a parallel with the human brain, which takes years to develop intelligence through experiences and sensory inputs, we see that simply inundating a less powerful computer with unstructured data won't make it intelligent overnight. The key solution lies in treating AI as a learner and leveraging the wealth of knowledge in a developed human brain to teach it.

In our project, we utilize passive BCIs to allow AI to learn directly from the brain, understanding values, interpretations, goals, intentions, and even skills. This approach addresses the challenge of transforming unstructured data into meaningful intelligence.

The project employs three proven Proof of Concept mechanisms, showcased in our PNAS paper, demonstrating that AI can learn and perform tasks better through reinforcement learning based on reward functions derived from the human brain. This method ensures that AI gains a deeper understanding of the world, making it more effective in its functions than if it were to learn without insights from the human brain.

(TZ): Zander Labs is currently in a phase of significant growth, with plans to expand its team to around 80 people working on the project. The aim is to establish the company as a prominent DeepTech player in Europe, contributing to the global development of AI and new technologies.

The goal is not to replicate what larger entities like OpenAI are doing but to offer an alternative perspective shaped by European values, views, and legal constraints.

Zander Labs specifically focuses on AI, leveraging its NeuroTech background to pioneer new types of AI technologies. While these innovations might be compatible with existing technologies, the emphasis is on providing unique contributions, such as neuroadaptive AI, which addresses aspects like empathy that are not widely explored in the current AI landscape. The company envisions collaborative partnerships with major players in the U.S., working as equal contributors to advance the field collectively.