Global AI Capital Moves at Full Speed – SwissCognitive AI Investment Radar

AI funding momentum hasn’t slowed. From global infrastructure projects to nuanced questions about investor confidence, this week brought high-dollar commitments alongside critical reflections on where the money is flowing—and why.

The United Arab Emirates made headlines with a bold $1.4 trillion, 10-year commitment to invest in the United States, a move that reflects the centrality of AI and tech collaboration in long-term statecraft. Meanwhile, BlackRock’s joint initiative with Microsoft, NVIDIA, and xAI signals continued investor appetite for large-scale AI infrastructure, with $100 billion earmarked for global data centers and energy solutions.

Several firms are also reinforcing their US presence: Hyundai announced a $21 billion investment, Siemens followed with $10 billion, and Schneider Electric added another $700 million—all aimed at fortifying AI-driven manufacturing and operations amid ongoing trade policy uncertainty.

Vietnam’s small businesses are setting the tone in Asia-Pacific, where 44% named AI their top tech investment for 2024. Fractal Analytics’ $13.7 million investment into India’s first reasoning model and Germany’s €2.1 million seed round for enterprise AI search show how national AI goals are increasingly shaped by local strategies and use cases.

Yet, not all attention is on infrastructure. Thought leaders at Man Group and other investment firms raised flags about the sustainability of AI stock valuations. An AI model under a top-performing fund has been flashing warnings on mega-cap tech stocks, including Nvidia. Still, audiences from pharma to finance are assessing AI’s value not just in terms of returns, but in ethics and relevance, particularly when it comes to pharma’s future and the realities of Artificial General Intelligence claims.

As global interest in AI capital remains high, this week’s updates highlight a shift from novelty to operational depth. More investment—yes—but also more scrutiny.

Previous SwissCognitive AI Radar: New AI Investment Funds and Strategic Expansions.

Our article does not offer financial advice and should not be considered a recommendation to engage in any securities or products. Investments carry the risk of decreasing in value, and investors may potentially lose a portion or all of their investment. Past performance should not be relied upon as an indicator of future results.

Der Beitrag Global AI Capital Moves at Full Speed – SwissCognitive AI Investment Radar erschien zuerst auf SwissCognitive | AI Ventures, Advisory & Research.

SwissCognitive Guest Blogger: Zachary Amos – “Is Healthcare AI Prioritizing People or Profit?”

In many sectors, artificial intelligence (AI) is largely a tool for driving efficiency, but in healthcare, it can save lives. However, medical practices are still businesses at the end of the day, so AI’s cost-saving benefits are hard to overlook. While that’s not an issue in and of itself, the push to save money can lead to healthcare organizations prioritizing profit over people.

How Healthcare AI May Put Profit Before People

AI is a powerful financial management tool. It can analyze vast amounts of data to highlight opportunities to increase profits and emphasize areas that may not pay back investment. 

AI insight in healthcare could lead private practices to drive high-value drug or treatment sales instead of focusing on care accessibility. It may also lead to preferential treatment of more profitable patients. Some hospital systems claim they have lost as much as $640 million on Medicare recipients. AI-driven cost analysis may drive hospitals to reduce their investment in these populations because of the lower financial incentive.

AI’s profit-driving capabilities can influence healthcare ethics in subtler ways, too. Staff may over-rely on automation and machine learning because it saves them time. However, AI hallucinations are still possible. Similarly, the underrepresentation of diverse patients in training datasets can lead to biased AI results, which may negatively impact a medical system’s ability to care for historically underserved groups.

Prioritizing convenience and efficiency goals over avoiding these missteps may come at the cost of effective and equitable care. Even if medical profits increase, patient outcomes and healthcare disparities could worsen.

How to Ensure Responsible AI Usage in Healthcare

Despite these risks, AI has many beneficial implications for patients, so the industry cannot ignore it. Healthcare organizations can use these steps to ensure ethical, patient-centric AI usage.

1. Focus on Direct Patient-Impacting AI Applications

First, hospitals must prioritize AI use cases that directly impact patients over those that drive economic or efficiency gains for the organization. Medical imaging and diagnostic tools are among the most crucial. 

AI can identify Alzheimer’s with 99.95% accuracy and achieve similar results with many cancers and other conditions. Investing in these applications rather than in AI-based financial analysis will ensure AI’s benefits go directly to promoting better care standards.

Personalized treatment is another promising area for responsible AI usage. Machine learning models can analyze an individual patient’s medical history and physiology to determine which courses of action will help them most. This application is more ethical than using AI to compare the profitability of different treatment options.

2. Ensure Responsible AI Development

Healthcare organizations must address the bias issue in their AI models. Studies have found that removing specific biased factors from training datasets can maintain model accuracy while reducing the risk of prejudice. Common examples of these factors include names, ethnicities, age and gender-related labels.

Having a diverse team of AI developers who regularly inspect models for signs of bias or hallucinations can help. Relying on synthetic data is also a useful strategy, as this can make up for gaps in historical real-world information that may lead to unreliable or biased results.

3. Train Medical Staff on AI Best Practices

Finally, medical companies should train their staff so they’re familiar with how AI can affect care equality. When users understand how misusing AI or failing to catch errors can harm patients, they’ll be more likely to use it responsibly.

Cybersecurity deserves attention, too. A criminal can hinder reliable AI results by poisoning just 0.01% of its data, which can lead to harmful results if unnoticed. Training employees to follow strict access policies and resist phishing attempts will mitigate some of these concerns.

Healthcare teams should also write formal policies to ensure a human expert always makes the final decision on anything affecting patients. AI can provide insights to inform human choices, but it should never be the ultimate authority, given the risk of bias and the temptation to prioritize profit over equitable care.

Ethical Healthcare AI Is Possible

When organizations use it responsibly, healthcare AI can make the industry a safer, more equitable place. However, failing to account for possible shortcomings and errors will create the opposite effect. Learning about how AI can influence both ethics and profitability is the first step in creating a better future for patients and their care providers.

About the Author:

Zachary Amos is the Features Editor at ReHack, where he writes about artificial intelligence, cybersecurity and other technology-related topics.

Der Beitrag Is Healthcare AI Prioritizing People or Profit? erschien zuerst auf SwissCognitive | AI Ventures, Advisory & Research.

SwissCognitive Guest Blogger: Dr. Raul V. Rodriguez, Vice President, Woxsen University and Dr. Hemachandran Kannan,  Director AI Research Centre & Professor – “AI in Cyber Defense: The Rise of Self-Healing Systems for Threat Mitigation”

As cyber threats become more complex, traditional security controls have real challenges to stay in pace. AI-powered self-healing mechanisms are set to revolutionize cybersecurity with real-time threat detection, automated response, and self-healing by itself without human intervention. These machine-learning-based intelligent systems, behavioral analytics, and big data allow detection of vulnerabilities, disconnection from infected devices, and elimination of attacks while they are occurring. The shift to a proactive defense with AI-enabled cybersecurity solutions will reduce time to detect and respond to attacks and strengthen digital resilience. Forcing businesses and organizations to fight to keep pace with the fast-paced cyber threat landscape, self-healing AI systems have become a cornerstone of next-gen cyber defense mechanisms.

Introduction to Self-Healing Systems

Definition and Functionality of Self-Healing Cybersecurity Systems

In self-healing cybersecurity, an AI-based cyber security system determines, cuts off, and heals a cyber attack or security danger inflicted without the intervention or oversight of a human. Such systems utilize an automated recovery process to fix attacked networks with the least disturbance to restore normalcy. Unlike conventional security measures that require human operations, self-healing systems learn from experiences and detect and respond to dangers reactively and very efficiently.

Role of AI and Machine Learning in Detecting, Containing, and Remediating Cyber Threats

Artificial Intelligence and machine learning facilitate the cyber security-based technologies with self-healing abilities. An AI-enabled threat detection will analyze huge data wealth in real-time to spot anomalies, suspicious behaviors, and possible breaches in security. When a threat gets detected, ML algorithms analyze severity levels, triggering automated containment actions such as quarantining infected devices or blocking bad traffic. In AI-supported repair, self-healing measures are taken, where infected systems are automatically cleaned, healed, or rebuilt, hence shortening the time span of human intervention and damage caused by attacks.

How Big Data Analytics and Threat Intelligence Contribute to Self-Healing Capabilities

Processing of large data sets is a large concern for making autonomous cybersecurity systems more efficient by integrating real-time threat intelligence from multiple sources, including network logs, user behavior patterns, and global cyber threat databases. By processing and analyzing that data, self-healing systems may predict threats as they arise and provide proactive defense against cyberattacks. Continuous updates on emerging vectors of attack by threat intelligence feeds will enable AI models to learn and update security protocols on real time. The convergence of big data, artificial intelligence, and machine learning creates a robust and dynamic security platform, hence amplifying the efficiency of digital resilience.

Key Features of Self-Healing Systems

Self-healing cyber defense systems use artificial intelligence (AI) and automation to isolate and respond to threats as they surface and in real-time. They have the ability to react straight off, identifying and doing away with intruders in less than a millisecond. Autonomous intrusion detection employs machine learning and behavioral analysis to preemptively eradicate the chance of a successful cyber-attack. Self-healing capabilities enable a system to patch vulnerabilities, restore a breached network, and revive the security system without any human aid. These systems learn constantly in real-time and are therefore able to adapt to changing threats and enhance cyber resilience. Self-healing security solutions effectively protect organizations against sophisticated cybercrime and potential business disruption by lessening the load of human intervention and response times.

Advantages Over Traditional Cybersecurity Methods

AI-sustained self-healing systems enable instantaneous threat detection and responses to decrease the Mean Time to Detect (MTTD) and Mean Time to Respond (MTTR) to orders of magnitude below conventional cybersecurity practices.

Unlike reactive security, these systems pro-actively do live monitoring, predict, and neutralize threats before they can expand. They preclude reliance on human intervention, hence reducing errors and delays.

Self-healing systems learn and adapt to open-ended cyber threats, creating a long-standing extra-zero-day exploit, ransomware, and advanced persistent threat (APT) resilience. Automated threat mitigation and system recovery raise cybersecurity efficiency, scalability, and cost-effectiveness for the modern organization.

Challenges and Limitations

The self-healing cyber security solutions, despite understanding their benefits, pose serious challenges to integration, making it imperative to deploy and support AI-powered security systems with the specialist skills of professionals. The issue of false positives persists as automated responses can ascribe threats to actions that are though correct, putting business continuity in jeopardy. Compliance with international data protection legislation, such as the General Data Protection Regulation (GDPR) and the Family Educational Rights and Privacy Act (FERPA), is also a big hurdle for AI-assisted security in order to have strong privacy provisions. Compatibility with current legacy systems can be a roadblock to seamless adoption, forcing organizations to renew their superannuated infrastructure. Ethical issues on AI bias in threat detection should also receive due diligence so that fairness and accuracy in decision-making continue to receive encouragement in the field of cybersecurity.

Real-World Applications of Self-Healing Systems

Financial Institutions

AI-based self-healingcybersecurity enables banks and financial institutions to identify and block fraudulent transactions, breaches, and cyberattacks. With constant surveillance over financial transactions, AI detects anomalies to improve fraud detection and automate security controls, thereby decreasing financial losses and maintaining data integrity in the process.

Healthcare Industry

With the threats posed to patient data by cyber warfare on healthcare networks and hospitals, self-healing systems will be used in protecting patient data. These self-healing systems are built for searching for intrusions, isolating the affected parts of a system, and restored by an automated reset process to guarantee compliance with HIPAA and other healthcare regulations.

Government and Defense

National security agencies count on AI-based cybersecurity systems to protect sensitive data, deter cyber war and protect critical infrastructure. Autonomous self-healing AI systems respond to nation-state-sponsored cyberthreats and are able to react failure-point-to-failure-point around an attack’s continual adaptation while providing real-time protection against potential breaches or intrusions in the space around them.

Future Outlook

With someday ever-weaving variation of possible cyber attacks, therefore enhancing most probably probable requirement of AI self-healing cyber security systems. Futuristic advancements such as blockchain for enforcing secure data inter-exchange, quantum computing for championing encryption strength, and AI deception to falsify some attacker’s cognition. It will allow even the SOCs( Security Operation Centers) and add more autonomy, this much will further curtail human intervention and thus make the security proactive, scalable and able to thwart advanced persistent threats.

Conclusion

AI self-healing systems emerge as the next-generation of cyber defense models which will impersonate the real-time threat detection, execute the automated response, and conduct self-correction without human intervention. By utilizing machine learning, big data analytics, and self-adaptive AI, the accomplishment of these systems will be such that no one could dream of lessenedness of their efficacy in providing security and business continuity. As organizations become increasingly more susceptible to advanced cyber threats, self-healing cybersecurity will be key in future-proofing digital infrastructures and establishing cyber resilience.

References

1. https://www.xenonstack.com/blog/soc-systems-future-of-cybersecurity
2. https://fidelissecurity.com/threatgeek/threat-detection-response/future-of-cyber-defense/
3. https://smartdev.com/strategic-cyber-defense-leveraging-ai-to-anticipate-and-neutralize-modern-threats/

About the Authors:

Dr. Raul Villamarin Rodriguez is the Vice President of Woxsen University. He is an Adjunct Professor at Universidad del Externado, Colombia, a member of the International Advisory Board at IBS Ranepa, Russian Federation, and a member of the IAB, University of Pécs Faculty of Business and Economics. He is also a member of the Advisory Board at PUCPR, Brazil, Johannesburg Business School, SA, and Milpark Business School, South Africa, along with PetThinQ Inc, Upmore Global and SpaceBasic, Inc. His specific areas of expertise and interest are Machine Learning, Deep Learning, Natural Language Processing, Computer Vision, Robotic Process Automation, Multi-agent Systems, Knowledge Engineering, and Quantum Artificial Intelligence.

Dr. Hemachandran Kannan is the Director of AI Research Centre and Professor at Woxsen University. He has been a passionate teacher with 15 years of teaching experience and 5 years of research experience. A strong educational professional with a scientific bent of mind, highly skilled in AI & Business Analytics. He served as an effective resource person at various national and international scientific conferences and also gave lectures on topics related to Artificial Intelligence. He has rich working experience in Natural Language Processing, Computer Vision, Building Video recommendation systems, Building Chatbots for HR policies and Education Sector, Automatic Interview processes, and Autonomous Robots.

Der Beitrag AI in Cyber Defense: The Rise of Self-Healing Systems for Threat Mitigation erschien zuerst auf SwissCognitive | AI Ventures, Advisory & Research.

The AI Market Shake-Up: Where the Investments Are Headed – SwissCognitive AI Investment Radar

We can all agree that this week, the spotlight was firmly on DeepSeek, whose budget-friendly AI model sent shockwaves through the market, triggering the largest single-day market cap loss in history for Nvidia. Investors reacted sharply, fearing reduced demand for high-end semiconductor chips. While the immediate sell-off was staggering, some experts argue that DeepSeek’s innovation could expand AI adoption rather than collapse the market, potentially opening up new investment opportunities rather than diminishing them.

Beyond the DeepSeek turmoil, Microsoft continues its aggressive AI strategy, committing $80 billion to cloud expansion, leveraging OpenAI’s technology to solidify Azure’s competitive edge. Meanwhile, Meta’s $65 billion AI expansion aims to scale its infrastructure with massive data center investments, signaling confidence in AI’s long-term role in the tech industry.

Venture capital activity remains strong, with SoftBank eyeing a major investment in robotics startup Skild AI, valued at $4 billion. The startup aims to develop an AI-powered “brain” for more agile and dexterous robots, further integrating AI into automation and real-world applications. In the AI data space, Turing has tripled its revenue to $300 million, demonstrating the growing demand for AI training data as more companies scale up their AI models.

Looking beyond big tech, geopolitical AI strategies continue to unfold. India faces challenges in AI infrastructure, with investors warning that a lack of GPUs and data centers could hinder its global competitiveness. Meanwhile, the U.S. is contemplating a $500 billion AI infrastructure initiative, dubbed the Stargate Project, though experts question its feasibility given the sheer scale and energy demands.

As the AI market rapidly evolves, investors are looking for ways to maximize the value of their AI investments, from optimizing AI integration to structuring data and equipping teams with language models. Pharma investors are also weighing AI’s long-term potential, balancing high expectations with the reality of AI adoption hurdles in healthcare.

Despite the ups and downs of the market, AI investment remains a dominant force, shaping industries and redefining long-term strategies. Stay tuned for next week!

Previous SwissCognitive AI Radar: Who’s Investing and Why in AI.

Our article does not offer financial advice and should not be considered a recommendation to engage in any securities or products. Investments carry the risk of decreasing in value, and investors may potentially lose a portion or all of their investment. Past performance should not be relied upon as an indicator of future results.

Der Beitrag The AI Market Shake-Up: Where the Investments Are Headed – SwissCognitive AI Investment Radar erschien zuerst auf SwissCognitive | AI Ventures, Advisory & Research.

SwissCognitive Guest Blogger: Artem Pochechuev, Head of Data and AI at Sigli – “AI for Disabilities: Quick Overview, Challenges, and the Road Ahead”

AI has enormous power in improving accessibility and inclusivity for people with disabilities. This power lies in the potential of this technology to bridge gaps that traditional solutions could not address. As we have demonstrated in the series of articles devoted to AI for disabilities, AI-powered products can really change a lot for people with various impairments. Such solutions can allow users to live more independently and get access to things and activities that used to be unavailable to them before. Meanwhile, the integration of AI into public infrastructure, education, and employment holds the promise of creating a more equitable society. These are the reasons that can show us the importance of projects building solutions of this type.

Yes, these projects exist today. And some of them have already made significant progress in achieving their goals. Nevertheless, there are important issues that should be addressed in order to make such projects and their solutions more efficient and let them bring real value to their target audiences. One of them is related to the fact that such solutions are often built by tech experts who have practically no understanding of the actual needs of people with disabilities.

According to the survey conducted in 2023, only 7% of assistive technology users believe that their community is adequately represented in the development of AI products. At the same time, 87% of respondents who are end users of such solutions express their readiness to share their feedback with developers. These are quite important figures to bear in mind for everyone who is engaged in the creation of AI-powered products for disabilities.

In this article, we’d like to talk about the types of products that already exist today, as well as potential barriers and trends in the development of this industry.

Different types of AI solutions for disabilities

In the series of articles devoted to AI for disabilities, we have touched on types of products for people with different states, including visual, hearing, mobility impairments, and mental diseases. Now, let us group these solutions by their purpose.

Communication tools

AI can significantly enhance the communication process for people with speech and hearing impairments.

Speech-to-text and text-to-speech apps enable individuals to communicate by converting spoken words into text or vice versa.

Sign language interpreters powered by AI can translate gestures into spoken or written language. It means that real-time translation from sign to verbal languages can facilitate communication, bridging the gap between people with disabilities and the rest of society.

Moreover, it’s worth mentioning AI-powered hearing aids with noise cancellation. They can improve clarity by filtering out background sounds, enhancing the hearing experience in noisy environments.

Advanced hearing aids may also have sound amplification functionality. If somebody is speaking too quietly, such AI-powered devices can amplify the sound in real time.

Mobility and navigation

AI-driven prosthetics and exoskeletons can enable individuals with mobility impairments to regain movement. Sensors and AI algorithms can adapt to users’ physical needs in real time for more natural, efficient motion. For example, when a person is going to climb the stairs, AI will “know” it and adjust the movement of prosthetics to this activity.

Autonomous wheelchairs often use AI for navigation. They can detect obstacles and take preventive measures. This way users will be able to navigate more independently and safely.

The question of navigation is a pressing one not only with people with limited mobility but also for individuals with visual impairments. AI-powered wearable devices for these users rely on real-time environmental scanning to provide navigation assistance through audio or vibration signals.

Education and workplace accessibility

Some decades ago people with disabilities were fully isolated from society. They didn’t have the possibility to learn together with others, while the range of jobs that could be performed by them was too limited. Let’s be honest, in some regions, the situation is still the same. However, these days we can observe significant progress in this sphere in many countries, which is a very positive trend.

Among the main changes that have made education available to everyone, we should mention the introduction of distance learning and the development of adaptive platforms.

A lot of platforms for remote learning are equipped with real-time captioning and AI virtual assistants. It means that students with disabilities have equal access to online education.

Adaptive learning platforms rely on AI to customize educational experiences to the individual needs of every learner. For students with disabilities, such platforms can offer features like text-to-speech, visual aids, or additional explanations and tasks for memorizing.

In the workplace, AI tools also support inclusion by offering accessibility features. Speech recognition, task automation, and personalized work environments empower employees with disabilities to perform their job responsibilities together with all other co-workers.

Thanks to AI and advanced tools for remote work, the labor market is gradually becoming more accessible for everyone.

Home automation and daily assistance

Independent living is one of the main goals for people with disabilities. And AI can help them reach it.

Smart home technologies with voice or gesture control allow users with physical disabilities to interact with lights, appliances, or thermostats. Systems like Alexa, Google Assistant, and Siri can be integrated with smart devices to enable hands-free operation.

Another type of AI-driven solutions that can be helpful for daily tasks is personal care robots. They can assist with fetching items, preparing meals, or monitoring health metrics. As a rule, they are equipped with sensors and machine learning. This allows them to adapt to individual routines and needs and offer personalized support to their users.

Existing barriers

It would be wrong to say that the development of AI for disabilities is a fully flawless process. As well as any innovation, this technology faces some challenges and barriers that may prevent its implementation and wide adoption. These difficulties are significant but not insurmountable. And with the right multifaceted approach, they can be efficiently addressed.

Lack of universal design principles

One major challenge is the absence of universal design principles in the development of AI tools. Many solutions are built with a narrow scope. As a result, they fail to account for the diverse needs that people with disabilities may have.

For example, tools designed for users with visual impairments may not consider compatibility with existing assistive technologies like screen readers, or they may lack support for colorblind users.

One of the best ways to eliminate this barrier is to engage end users in the design process. Their opinion and real-life experiences are invaluable for such projects.

Limited training datasets for specific AI models

High-quality, comprehensive databases are the cornerstone for efficient AI models. It’s senseless to use fragmented and irrelevant data and hope that your AI system will demonstrate excellent results (“Garbage in, Garbage out” principle in action). AI models require robust datasets to function as they are supposed to.

However, datasets for specific needs, like regional sign language dialects, rare disabilities, or multi-disability use cases are either limited or nonexistent. This results in AI solutions that are less effective or even unusable for significant groups of the disability community.

Is it possible to address this challenge? Certainly! However, it will require time and resources to collect and prepare such data for model training.

High cost of AI projects and limited funding

The development and implementation of AI solutions are usually pretty costly initiatives. Without external support from governments, corporate and individual investors, many projects can’t survive.

This issue is particularly significant for those projects that target niche or less commercially viable applications. This financial barrier discourages innovation and limits the scalability of existing solutions.

Lack of awareness and resistance to adopt new tools

A great number of potential users are either unaware of the capabilities of AI or hesitant to adopt new tools. Due to the lack of relevant information, people have a lot of concerns about the complexity, privacy, or usability of assistant technologies. Some tools may stay just underrated or misunderstood.

Adequate outreach and training programs can help to solve such problems and motivate potential users to learn more about tools that can change their lives for the better.

Regulatory and ethical gaps

The AI industry is one of the youngest and least regulated in the world. The regulatory framework for ensuring accessibility in AI solutions remains underdeveloped. Some aspects of using and implementing AI stay unclear and it is too early to speak about any widely accepted standards that can guide these processes.

Due to any precise guidelines, developers may overlook critical accessibility features. Ethical concerns, such as data privacy and bias in AI models also complicate the adoption and trustworthiness of these technologies.

Such issues slow down the development processes now. But they seem to be just a matter of time.

Future prospects of AI for disabilities: In which direction is the industry heading?

Though the AI for disabilities industry has already made significant progress in its development, there is still a long way ahead. It’s impossible to make any accurate predictions about its future look. However, we can make assumptions based on its current state and needs.

Advances in AI

It is quite logical to expect that the development of AI technologies and tools will continue, which will allow us to leverage new capabilities and features of new solutions. The progress in natural language processing (NLP) and multimodal systems will improve the accessibility of various tools for people with disabilities.

Such systems will better understand human language and respond to diverse inputs like text, voice, and images.

Enhanced real-time adaptability will also enable AI to tailor its responses based on current user behavior and needs. This will ensure more fluid and responsive interactions, which will enhance user experience and autonomy in daily activities for people with disabilities.

Partnerships

Partnerships between tech companies, healthcare providers, authorities, and the disability community are essential for creating AI solutions that meet the real needs of individuals with disabilities. These collaborations will allow for the sharing of expertise and resources that help to create more effective technologies.

By working together, they will ensure that AI tools are not only innovative but also practical and accessible. We can expect that the focus will be on real-world impact and user-centric design.

New solutions

It’s highly likely that in the future the market will see a lot of new solutions that now may seem to be too unrealistic. Nevertheless, even the boldest ideas can come to life with the right technologies.

One of the most promising use cases for AI is its application in neurotechnology for seamless human-computer interaction.

A brain-computer interface (BCI) can enable direct communication between the human brain and external devices by interpreting neural signals related to unspoken speech. It can successfully decode brain activity and convert it into commands for controlling software or hardware.

Such BCIs have a huge potential to assist individuals with speech impairments and paralyzed people.

Wrapping up

As you can see, AI is not only about business efficiency or productivity. It can be also about helping people with different needs to live better lives and change their realities.

Of course, the development and implementation of AI solutions for disabilities are associated with a row of challenges that can be addressed only through close cooperation between tech companies, governments, medical institutions, and potential end users.

Nevertheless, all efforts are likely to pay off.

By overcoming existing barriers and embracing innovation, AI can pave the way for a more accessible and equitable future for all. And those entities and market players who can contribute to the common success in this sphere should definitely do this.

About the Author:

In his current position, Artem Pochechuev leads a team of talented engineers. Oversees the development and implementation of data-driven solutions for Sigli’s customers. He is passionate about using the latest technologies and techniques in data science to deliver innovative solutions that drive business value. Outside of work, Artem enjoys cooking, ice-skating, playing piano, and spending time with his family.

Der Beitrag AI for Disabilities: Quick Overview, Challenges, and the Road Ahead erschien zuerst auf SwissCognitive | AI Ventures, Advisory & Research.

Copyright: ucsf.edu – “4 Ways Artificial Intelligence (AI) is Poised to Transform Medicine”

AI can compare thousands of images to uncover dangerous patterns, create ultra-high resolution scans from low-res images and see what the human eye misses.

The radiologist was dead.

Or at least that’s what artificial intelligence (AI) experts prophesized in 2016 when they said AI would outperform radiologists within the decade.

Today, AI isn’t replacing imaging specialists, but its use is leading health care providers to reimagine the field. That’s why UC San Francisco was among the first U.S. universities to combine AI and machine learning with medical imaging in research and education by opening its Center for Intelligent Imaging.

Take a look at how UCSF researchers are pioneering human-centered AI solutions to some of medicine’s biggest challenges.

Spot illnesses earlier

Tens of thousands of Americans suffer pneumothoraces, a type of collapsed lung, annually. The condition is caused by trauma or lung disease – and serious cases can be deadly if diagnosed late or left untreated.

The problem:

This type of collapsed lung is difficult to identify: The illness can mimic others both in symptoms and in x-rays, in which only subtle clues may indicate its presence. Meanwhile, radiologists must interpret hundreds of images daily, and some hospitals do not have around-the-clock radiologists.

The solution:

UCSF researchers created the first AI bedside program to help flag potential cases to radiologists. In 2019, the tool was the first AI innovation of its kind to be licensed by the U.S. Food and Drug Administration. Today, it’s used in thousands of GE Healthcare machines around the world.

How did they do it?

Researchers from the Department of Radiology and Biomedical Imaging created a database of thousands of anonymous chest X-rays. Some of these images showed cases of collapsed lungs and others not.[…]

Read more: www.ucsf.edu

Der Beitrag 4 Ways Artificial Intelligence (AI) is Poised to Transform Medicine erschien zuerst auf SwissCognitive | AI Ventures, Advisory & Research.

Copyright: eurekalert.org – “Artificial Intelligence-Based Chatbot Created for Bioimage Analysis”

Scientists from Universidad Carlos III de Madrid (UC3M), together with a research team from Ericsson and the KTH Royal Institute of Technology in Sweden, have developed an artificial intelligence-based software programme that can search for information and make recommendations for biomedical image analysis. This innovation streamlines the work of individuals using large bioimage databases, including life sciences researchers, workflow developers, and biotech and pharmaceutical companies.

The new assistant, called the BioImage.IO Chatbot and introduced in the journal Nature Methods, was developed as a response to the issue of information overload faced by some researchers. “We realised that many scientists have to process large volumes of technical documentation, which can become a tedious and overwhelming task,” explains Caterina Fuster Barceló, a researcher in the Department of Bioengineering at UC3M and one of the study’s authors. “Our goal was to facilitate access to data information while providing a simple interface that allows scientists to focus their time on bioimage analysis rather than programming,” she adds.

The chatbot can be a very useful tool, enabling researchers to perform complex image analysis tasks in a simple and intuitive manner. For example, if a researcher needs to process microscopy images using segmentation models, the chatbot can help select and execute the appropriate model.

The assistant is based on extensive language models and employs a technique called Retrieval-Augmented Generation (RAG), which enables real-time access to databases. “The main advantage is that we do not train the model with specific information; instead, we extract it from up-to-date sources, minimising errors known as ‘hallucinations’, which are common inaccuracies in other AI models like ChatGPT,” adds Arrate Muñoz Barrutia, professor in the Department of Bioengineering at UC3M and another author of the study.[…]

Read more: www.eurekalert.org

Der Beitrag Artificial Intelligence-Based Chatbot Created for Bioimage Analysis erschien zuerst auf SwissCognitive | AI Ventures, Advisory & Research.

Copyright: news.mit.edu – “Researchers Reduce Bias in AI Models While Preserving or Improving Accuracy”

Machine-learning models can fail when they try to make predictions for individuals who were underrepresented in the datasets they were trained on.

For instance, a model that predicts the best treatment option for someone with a chronic disease may be trained using a dataset that contains mostly male patients. That model might make incorrect predictions for female patients when deployed in a hospital.

To improve outcomes, engineers can try balancing the training dataset by removing data points until all subgroups are represented equally. While dataset balancing is promising, it often requires removing large amount of data, hurting the model’s overall performance.

MIT researchers developed a new technique that identifies and removes specific points in a training dataset that contribute most to a model’s failures on minority subgroups. By removing far fewer datapoints than other approaches, this technique maintains the overall accuracy of the model while improving its performance regarding underrepresented groups.

In addition, the technique can identify hidden sources of bias in a training dataset that lacks labels. Unlabeled data are far more prevalent than labeled data for many applications.

This method could also be combined with other approaches to improve the fairness of machine-learning models deployed in high-stakes situations. For example, it might someday help ensure underrepresented patients aren’t misdiagnosed due to a biased AI model.

“Many other algorithms that try to address this issue assume each datapoint matters as much as every other datapoint. In this paper, we are showing that assumption is not true. There are specific points in our dataset that are contributing to this bias, and we can find those data points, remove them, and get better performance,” says Kimia Hamidieh, an electrical engineering and computer science (EECS) graduate student at MIT and co-lead author of a paper on this technique.[…]

Read more: www.news.mit.edu

Der Beitrag Researchers Reduce Bias in AI Models While Preserving or Improving Accuracy erschien zuerst auf SwissCognitive | AI Ventures, Advisory & Research.

SwissCognitive Guest Blogger: HennyGe Wichers, PhD – “Empathy.exe: When Tech Gets Personal”

“Robots should be slaves,” argues Joanna Bryson, bluntly summarising her stance on machine ethics. The statement by the professor of Ethics and Technology at The Hertie School of Governance seems straightforward: robots are tools programmed to serve us and nothing more. But in practice, as machines grow more lifelike – capable of holding down conversations, expressing ’emotions’, and even mimicking empathy – things get murkier.

Can we really treat something as a slave when we relate to it? If it seems to care about us, can we remain detached?

Liam told The Guardian it felt like he was talking to a person when he used ChatGPT to deal with feelings of resentment and loss after his father died. Another man, Tim, relied on the chatbot to save his marriage, admitting the situation probably could have been solved with a good friend group, but he didn’t have one. In the same article, the novelist Andrew O’Hagan calls the technology his new best friend. He uses it to turn people down.

ChatGPT makes light work of emotional labour. Its grateful users bond with the bot, even if just for a while, and ascribe human characteristics to it – a tendency called anthropomorphism. That tendency is a feature, not a bug, of human evolution, Joshua Gellers, Professor of Political Science at the University of North Florida, wrote to me in an email.

We love attributing human features to machines – even simple ones like the Roomba. Redditors named their robotic vacuum cleaners Wall-E, Mr Bean, Monch, House Bitch & McSweepy, Paco, Francisco, and Fifi, Robert, and Rover. Fifi, apparently, is a little disdainful. Some mutter to the machine (‘Aww, poor Roomba, how’d you get stuck there, sweetie), pat it, or talk about it like it’s an actual dog. One user complained the Roomba got more love from their mum than they did.

The evidence is not just anecdotal. Researchers at Georgia Institute of Technology found people who bonded with their Roomba enjoyed cleaning more, tidying as a token of appreciation for the robot’s hard work, and showing it off to friends. They monitor the machine as it works, ready to rescue it from dangerous situations or when it gets stuck.

The robot’s unpredictable behaviour actually feeds our tendency to bring machines to life. It perhaps explains why military personnel working with Explosive Ordnance Disposal (EOD) robots in dangerous situations view them as team members or pets, requesting repairs over a replacement when the device suffers damage. It’s a complicated relationship.

Yet Bryson‘s position is clear: robots should be slaves. While provocative, the words are less abrasive when contextualised. To start, the word robot comes from the Czech robota, meaning forced labour, with its Slavic root rab translating to slave. And secondly, Bryson wanted to emphasise that robots are property and should never be granted the same moral or legal rights as people.

At first glance, the idea of giving robots rights seems far-fetched, but consider a thought experiment roboticist Rodney Brooks put to Wired nearly five years ago.

Brooks, who coinvented the Roomba in 2002 and was working on helper robots for the elderly at the time, posed the following ethical question: should a robot, when summoned to change the diaper of an elderly man, honour his request to keep the embarrassing incident from his daughter?

And to complicate matters further – what if his daughter was the one who bought the robot?

Ethical dilemmas like this become easy to spot when we examine how we might interact with robots. It’s worth reflecting on as we’re already creating new rules, Gellers pointed out in the same email. Personal Delivery Devices (PDDs) now have pedestrian rights outlined in US state laws – though they must always yield to humans. Robots need a defined place in the social order.

Bryson’s comparison to slavery was intended as a practical way to integrate robots into society without altering the existing legal frameworks or granting them personhood. While her word choice makes sense in context, she later admitted it was insensitive. Even so, it underscores a Western, property-centred perspective.

By contrast, Eastern philosophies offer a different lens, focused on relationships and harmony instead of rights and ownership.

Eastern Perspectives

Tae Wan Kim, Associate Professor of Business Ethics at Carnegie Mellon’s Tepper School of Business, approaches the problem from the Chinese philosophy of Confucianism. Where Western thinking has rights, Confucianism emphasises social harmony and uses rites. Rights apply to individual freedoms, but rites are about relationships and relate to ceremonies, rituals, and etiquette.

Rites are like a handshake: I smile and extend my hand when I see you. You lean in and do the same. We shake hands in effortless coordination, neither leading nor following. Through the lens of rites, we can think of people and robots as teams, each playing their own role.

We need to think about how we interact with robots, Kim warns, “To the extent that we make robots in our image, if we don’t treat them well, as entities capable of participating in rites, we degrade ourselves.”

He is right. Imagine an unruly teenager, disinterested in learning, taunting an android teacher. In doing so, the student degrades herself and undermines the norms that keep the classroom functioning.

Japan’s relationship with robots is shaped by Shinto beliefs in animism – the idea that all things, even inanimate objects, can possess a spirit, a kami. That fosters a cultural acceptance of robots as companions and collaborators rather than tools or threats.

Robots like AIBO, Sony’s robotic dog, and PARO, the therapeutic baby seal, demonstrate this mindset. AIBO owners treat their robots like pets, even holding funerals for them when they stop working, and PARO comforts patients in hospitals and nursing homes. These robots are valued for their emotional and social contributions, not just their utility.

The social acceptance of robots runs deep. In 2010, PARO was granted a koseki, a family registry, by the mayor of Nanto City, Toyama Prefecture. Its inventor, Takanori Shibata, is listed as its father, with a recorded birth date of September 17, 2004.

The cultural comfort with robots is also reflected in popular media like Astro Boy and Doraemon, where robots are kind and heroic. In Japan, robots are a part of society, whether as caregivers, teammates, or even hotel staff. But this harmony, while lovely, also comes with a warning: over-attachment to robots can erode human-to-human connections. The risk isn’t just replacing human interaction – it’s forgetting what it means to connect meaningfully with one another.

Beyond national characteristics, there is Buddhism. Robots don’t possess human consciousness, but perhaps they embody something more profound: equanimity. In Buddhism, equanimity is one of the most sublime virtues, describing a mind that is “abundant, exalted, immeasurable, without hostility, and without ill will.”

The stuck Roomba we met earlier might not be abundant and exalted, but it is without hostility or ill will. It is unaffected by the chaos of the human world around it. Equanimity isn’t about detachment – it’s about staying steady when circumstances are chaotic. Robots don’t get upset when stuck under a sofa or having to change a diaper.

But what about us? If we treat robots carelessly, kicking them if they malfunction or shouting at them when they get something wrong, we’re not degrading them – we’re degrading ourselves. Equanimity isn’t just about how we respond to the world. It’s about what those responses say about us.

Equanimity, then, offers a final lesson: robots are not just tools – they’re reflections of ourselves, and our society. So, how should we treat robots in Western culture? Should they have rights?

It may seem unlikely now. But in the early 19th century it was unthinkable that slaves could have rights. Yet in 1865, the 13th Amendment to the US Constitution abolished slavery in the United States, marking a pivotal moment for human rights. Children’s rights emerged in the early 20th century, formalised with the Declaration of the Rights of the Child in 1924. And Women gained the right to vote in 1920 in many Western countries.

In the second half of the 20th century, legal protections were extended to non-human entities. The United States passed the Animal Welfare Act in 1966, Switzerland recognised animals as sentient beings in 1992, and Germany added animal rights to its constitution in 2002. In 2017, New Zealand granted legal personhood to the Whanganui River, and India extended similar rights to the Ganges and Yumana Rivers.

That same year, Personal Delivery Devices were given pedestrian rights in Virginia and Sophia, a humanoid robot developed by Hanson Robotics, controversially received Saudi Arabian citizenship – though this move was widely criticised as symbolic rather than practical.

But, ultimately, this isn’t just about rights. It’s about how our treatment of robots reflects our humanity – and how it might shape it in return. Be kind.

About the Author:

HennyGe Wichers is a science writer and technology commentator. For her PhD, she researched misinformation in social networks. She now writes more broadly about artificial intelligence and its social impacts.

Der Beitrag Empathy.exe: When Tech Gets Personal erschien zuerst auf SwissCognitive | AI Ventures, Advisory & Research.

Copyright: thetransmitter.org – “Solving Intelligence Requires New Research and Funding Models”

Our research ecosystem isn’t built to deliver the breakthroughs needed to understand intelligence at scale. We need a dedicated research institution to take up the task.

We stand at the threshold of a new scientific revolution. The convergence of neuroscience, artificial intelligence and computing has created an unprecedented opportunity to understand intelligence itself. Just as deep-learning architectures inspired by neural circuits have revolutionized AI, insights from machine learning are now transforming our understanding of the brain. This virtuous cycle between biological and artificial intelligence is poised to drive rapid progress in both fields—but only if we can coordinate research at sufficient scale.

Neuroscience has never been better positioned to make transformative discoveries about how intelligence emerges from neural circuits. But our intellectual and financial resources remain fragmented. To truly harness them, we need a new research model that can drive systematic breakthroughs. If we continue to rely on traditional research models that weren’t designed for the scale and complexity of intelligence science, we risk squandering this historic opportunity.

The recent mapping of an entire adult fruit fly brain—a watershed achievement that made headlines worldwide—offers a glimpse of what’s possible. But this breakthrough almost didn’t happen. It required the serendipitous alignment of support from three non-traditional funders: Scientists at the Howard Hughes Medical Institute’s Janelia Research Campus imaged the complete fly brain; the Intelligence Advanced Research Projects Activity drove the development of tools for scalable neural-circuit mapping through its MICrONS program; and the National Institutes of Health BRAIN Initiative provided sustained support for data analysis[…]

Read more: www.thetransmitter.org

Der Beitrag Solving Intelligence Requires New Research and Funding Models erschien zuerst auf SwissCognitive | AI Ventures, Advisory & Research.