Lifeboat Foundation: Safeguarding Humanity
Toggle light / dark theme

Blog

As artificial intelligence models become increasingly advanced, electronics engineers have been trying to develop new hardware that is better suited for running these models, while also limiting power-consumption and boosting the speed at which they process data. Some of the most promising solutions designed to meet the needs of machine learning algorithms are platforms based on memristors.

Memristors, or memory resistors, are electrical components that can retain their resistance even in the absence of electrical power, adjusting their resistance based on the electrical charge passing through them. This means that they can simultaneously support both the storage and processing of information, which could be advantageous for running machine learning algorithms.

Memristor-based devices could be used to develop more compact and energy-efficient hardware for running AI models, including emerging distributed computing solutions referred to as edge computing systems. Despite their advantages, many existing -based platforms have been found to have notable limitations, adversely impacting their reliability and endurance.

Read more | >

Sound localization is one of the many learning tasks accomplished by the brain based on the binaural signals of the ears. Here, Wu et al demonstrate in-situ learning of sound localization function using a memristor array, with dramatic improvements in energy efficiency.

Read more | >

The human brain spends 30–50% of its waking hours engaged in mind-wandering (MW), a common phenomenon in which individuals either spontaneously or deliberately shift their attention away from external tasks to task-unrelated internal thoughts. Despite the significant amount of time dedicated to MW, its underlying reasons remain unexplained. Our pre-registered study investigates the potential adaptive aspects of MW, particularly its role in predictive processes measured by statistical learning. We simultaneously assessed visuomotor task performance as well as the capability to extract probabilistic information from the environment while assessing task focus (on-task vs. MW). We found that MW was associated with enhanced extraction of hidden, but predictable patterns.

Read more | >

A team of researchers from the University of Ottawa has developed innovative methods to enhance frequency conversion of terahertz (THz) waves in graphene-based structures, unlocking new potential for faster, more efficient technologies in wireless communication and signal processing.

THz waves, located in the far-infrared region of the electromagnetic spectrum, can be used to perform non-invasive imaging through for security and quality control applications. Additionally, these waves hold great promise for wireless communication.

Advances in THz , which can be used to change the frequency of electromagnetic waves, are essential for the development of high-speed and signal processing systems for 6G technologies and beyond.

Read more | >

Unlike traditional RLHFs, which only provide feedback after an assessment has been completed, pBCIs capture implicit, real-time information about the user’s cognitive and emotional state throughout the interaction. This allows the AI to access more comprehensive, multidimensional feedback, including intermediate decisions, judgments and thought processes. By observing brain activity when assessing situations, pBCIs provide a more comprehensive understanding of user needs and enable the AI to adapt more effectively and proactively.

By combining RLHF with pBCIs, we can elevate AI alignment to a new level—capturing richer, more meaningful information that enhances AI’s responsiveness, adaptability and effectiveness. This combination, called neuroadaptive RLHF, retains the standard RLHF approach but adds more detailed feedback through pBCIs in an implicit and unobtrusive way. Neuroadaptive RLHF allows us to create AI models that better understand and support the user, saving time and resources while providing a seamless experience.

The integration of RLHF with pBCIs presents both opportunities and challenges. Among the most pressing concerns are privacy and ethics, as pBCIs capture sensitive neural data. Ensuring proper consent, secure storage and ethical use of this data is critical to avoid misuse or breaches of trust.

Read more | >