AI understands many things but still flounders at human interaction
PARIS : Artificial intelligence continues to advance, yet this technology still struggles to grasp the complexity of human interactions. A recent US study reveals that, while AI excels at recognising objects or faces in still images, it remains ineffective at describing and interpreting social interactions in a moving scene.
The team led by Leyla Isik, professor of cognitive science at Johns Hopkins University, investigated how AI models understand social interactions.
To do this, the researchers designed a large-scale experiment involving over 350 AI models specialising in video, image or language. These AI tools were exposed to short, three-second video sequences illustrating various social situations.
At the same time, human participants were asked to rate the intensity of the interactions observed, according to several criteria, on a scale of 1-5. The aim was to compare human and AI interpretations, in order to identify differences in perception and better understand the current limits of algorithms in analysing our social behaviours.
The human participants were remarkably consistent in their assessments, demonstrating a detailed and shared understanding of social interactions. AI, on the other hand, struggled to match these judgements.
Models specialising in video proved particularly ineffective at accurately describing the scenes observed. Even models based on still images, although fed with several extracts from each video, struggled to determine whether the characters were communicating with each other.
As for language models, they fared a little better, especially when given descriptions written by humans, but remained far from the level of performance of human observers.
A 'blind spot'
For Isik, this proves a major obstacle to the integration of AI into real-world environments. 'AI for a self-driving car, for example, would need to recognise the intentions, goals, and actions of human drivers and pedestrians. You would want it to know which way a pedestrian is about to start walking, or whether two people are in conversation versus about to cross the street,' she explained.
'Any time you want an AI to interact with humans, you want it to be able to recognise what people are doing. I think this study sheds light on the fact that these systems can't right now.'
According to the researchers, this deficiency could be explained by the way in which AI neural networks are designed. These are mainly inspired by the regions of the human brain that process static images, whereas dynamic social scenes call on other brain areas.
This structural discrepancy could explain what the researchers suggest could be 'a blind spot in AI model development'.
Indeed, 'real life isn't static. We need AI to understand the story that is unfolding in a scene', said study co-author Kathy Garcia.
Ultimately, this research reveals a profound gap between the way humans and AI models perceive moving social scenes. Despite their computing power and ability to process vast quantities of data, machines are still unable to grasp the subtleties and implicit intentions underlying our social interactions.
Despite tremendous advances, artificial intelligence is still a long way from truly understanding exactly what goes on in human interactions.
Hashtags
Try Our AI Features
Explore what Daily8 AI can do for you:
Comments
No comments yet...
Related Articles
Borneo Post
4 hours ago
- Borneo Post
How AI will change all our lives forever
AI is already enabling personalised treatments, aiding in predictive diagnostics and improving patient care, and will see further inroads made in disease prevention and treatments. — Photo from ARTIFICIAL Intelligence (AI) has arrived in full force and will impact our future, transforming industries, the job market and our daily lives in ways that we haven't even thought of. The exciting possibilities that it poses challenge our imagination and give us hope for so many positive outcomes – in the fields of increased productivity, personalised healthcare and other innovative solutions. At the same time, AI has also raised our concerns about job displacement, data privacy and many ethical considerations. AI's integration into the various sectors of human endeavour, from manufacturing to finance, is already reshaping the way we work and live. Its continued development and presence (many call it 'intrusion') into our daily lives will require extremely careful management so that humans remain its masters to harness its myriad benefits and at the same time, mitigate its many potential risks and dangers. How will AI affect our future? AI-powered automation has already been used to streamline processes in various industries which has led to increased productivity and efficiency. In return, we have had reduced human errors and in cases of emergencies, autobots have been able to work around the clock without any breaks in between. The downside? Automation has meant the loss of jobs that previously were handled by humans. A drastic reduction in manpower needs spell doom for fast growing populations, especially those in India, China, and many Third World countries. There are five areas where I personally feel that AI will impact in a most positive way. They are revolutionising healthcare; transforming transportation; advancing agriculture; modernising finance; and reshaping retail trade. AI is already enabling personalised treatments, aiding in predictive diagnostics and improving patient care, and will see further inroads made in disease prevention and treatments. It is also driving improvements in autonomous vehicles, optimising traffic flow and advancing logistics in ways both accurately and expeditiously. In agriculture, AI has made strides in improving crop yields, managed environmental impact and optimized resource utilisation. It has also assisted in rendering ideas and statistical solutions for humans to further explore the new possibilities on gainful production systems. AI's use in the world of finance has assisted in fraud detection, algorithmic trading and in personalised financial advice. However, in this field, it has its detractors who feel that AI is also capable of intruding too much into our personal data and thus, the 'invasion of privacy' issue is still being hotly debated. Finally, in the field of retail trade, AI has been responsible for optimising e-commerce platforms, improving customer service and offering highly personalized shopping experiences. With the fast-growing trade online with speedy deliveries to homes and offices, this has meant more efficient systems and a faster delivery time frame for both supplier and customer. What about the impact of AI on the job market? Bill Gates was famously quoted having said in March this year (Source: CNBC) that most jobs would vanish in the 'Age of AI', but there would be three professions left that would outlast the machines: coders, energy experts and biologists; 'fields that demand creativity, real world problem-solving, and ethical judgement – all the key factors that AI is not able to fully replicate'. Gates continued to add that AI too would replace many doctors and teachers, and that humans would not be needed for most things! Geoffrey Hinton, one of the most respected minds in AI and often called the 'Godfather of AI', in a recent interview, had also sounded the alarm. Hinton said that AI could begin rewriting its own code within the next five years, and when that happened, it would become increasingly difficult for humans to understand, manage or control what these systems would be doing, or why they would be doing it. Scary, isn't it? As AI becomes more autonomous and capable of modifying itself, the risks multiply, from runaway learning systems to the development of goals misaligned with our human values and ethics. In the meantime, AI-powered automation will likely replace workers in routine and repetitive jobs, requiring a shift towards more creative and analytical roles. Individuals will need to adapt to the changing job market by acquiring new skills and embracing lifelong learning. Of most serious concern are the ethical and societal implications of what AI can do and will most likely be used for extensively – that of data privacy, security and regulation. All AI systems rely heavily or almost entirely on vast amounts of data, raising concerns about data privacy and security. There is also the fear that with AI algorithms, societal or guided biases input if aimed at biased targets can lead to unfair, discriminatory and even fatal outcomes. It is, therefore, crucial to develop and deploy AI systems responsibly, and for all users and those in positions to wield commands over them, to fully consider all these ethical implications to ensure fairness, balance and transparency. As AI becomes more pervasive by the day, it is up to our governments and organisations to establish appropriate regulations and governance frameworks to ensure its safe and ethical use. As the 'Godfather of AI, Hinton stresses the gravity of the urgent need for international cooperation and ethical safeguards. He has called for a global ban on autonomous AI weapons and a code of conduct to ensure that AI could not be exploited to spread misinformation, manipulate populations or develop dangerous levels of autonomy. AI, per se, holds immense potential to transform our future for the better, the greater good, but it also presents humankind significant challenges that require proactive management. We must address all the ethical concerns while fostering responsible development as we prepare our workforce for the changing landscape. With the right approach and a balanced mindset, we can use and harness the limitless power of AI to help us create a more prosperous, equitable, peaceful and harmonious future for us all. May God help us all if a day comes when the machines take total control. Amen. * The opinions expressed in this article are the columnist's own and do not reflect the view of the newspaper.
Barnama
4 hours ago
- Barnama
NACSA Finalising Malaysia's Cybersecurity Strategy 2025-2030 -- Digital Ministry
NACSA Finalising Malaysia's Cybersecurity Strategy 2025-2030 -- Digital Ministry KUALA LUMPUR, July 30 (Bernama) -- The National Cyber Security Agency (NACSA) is finalising Malaysia's Cyber Security Strategy (MCSS) 2025–2030 as a continuation of the MCSS 2020–2024, said the Ministry of Digital. The ministry said the MCSS will involve three key stakeholders: the government, businesses and the community. 'It also takes into account emerging technologies such as artificial intelligence (AI) and all related cybersecurity threats to ensure comprehensive protection of the country's digital ecosystem,' it posted on the Parliament's website. The Digital Ministry was responding to Datuk Seri Hishammuddin Hussein (BN-Sembrong) on national data security and sovereignty concerns regarding transparency in the training of artificial intelligence (AI) systems such as DeepSeek and the use of Huawei chips. The ministry added that it is also taking proactive steps to establish an AI security committee to review technical aspects and ensure compliance with security and ethical standards, including assessments of foreign technology. 'This committee will consist of experts from various fields in both the public and private sectors to strengthen governance and ensure that the development and use of AI in Malaysia remain safe, responsible, and aligned with national principles,' it said. Meanwhile, responding to a question from Muhammad Ismi Mat Taib (PN-Parit) regarding the progress of 5G coverage, the Ministry of Digital said that Digital Nasional Bhd (DNB) has developed the 5G network to reach 82.4 per cent coverage in populated areas (COPA), with a total of 7,489 5G sites as of June 30, 2025. Additionally, the ministry said the country's 5G penetration rate reached 80.4 per cent, with approximately 27.4 million active subscribers as of May 2025, further solidifying the country's position as a regional leader in digital connectivity and communication technology transformation. -- BERNAMA
The Sun
8 hours ago
- The Sun
UMS Integration moving up supply chain, expanding Penang operations
KUALA LUMPUR: UMS Integration Ltd is ramping up its push into the artificial intelligence (AI) and high‑performance computing supply chain, eyeing higher‑value precision components and expanding its Penang facilities as part of a five‑year growth roadmap following its landmark secondary listing on Bursa Malaysia today. CEO Luong Andy said the Singapore‑headquartered precision engineering specialist, which already counts two major semiconductor clients among its key customers, sees 'tremendous opportunities' in advanced packaging solutions critical for AI chips, a segment driving global capital expenditure in semiconductors. 'We are moving up the semiconductor value chain and see many opportunities, especially in AI and high‑performance computing, where we have the know‑how and capability to produce the precision components needed,' Luong told reporters after the listing ceremony. The company's shares opened at RM5.15, 3% above the reference price of RM5, and traded as high as RM5.39 in early deals, valuing UMS Integration at about RM3.7 billion. The counter, categorised under the Industrial Products and Services sector on Bursa Malaysia, closed at RM5.50 a share. Executive director Stanley Loh Meng Chong said UMS Integration is doubling down on Penang, where it operates its largest production plant, to support growing orders for advanced semiconductor equipment. The company recently acquired additional land in Penang and plans to begin construction next year, adding about 200,000 sq ft to reach 1 million sq ft in total manufacturing space within the next two to three years. 'Penang remains central to our volume manufacturing strategy. This expansion will support key customers' ramp‑up in operations, especially for advanced node manufacturing and packaging linked to AI chips,' Loh said, adding that Malaysia's accessibility and talent pool make it a preferred base for scaling production. UMS Integration's secondary listing on Bursa Malaysia was undertaken by way of introduction, meaning no new funds were raised. The company retains its primary listing on the Singapore Exchange, where it last traded at S$1.52. Luong said the dual‑listing strategy is aimed at enhancing liquidity and providing 'flexibility to tap different equity markets' for future fundraising if needed. 'This listing broadens our investor base and positions us better to access capital when opportunities arise,' he noted. For the first quarter ended March 31, 2025, UMS Integration posted S$57.7 million in revenue, up 7% year‑on‑year, with net profit edging up to S$10.1 million. The semiconductor segment contributed 84% of revenue, followed by aerospace at 11%. The group remains in a net cash position of S$81.4 million and continues its policy of quarterly dividends, paying S$0.01 per share in July. UMS Integration has historically been reliant on one major customer but is now balancing its portfolio with a second key client of comparable size. Loh said this diversification, coupled with demand from AI‑driven chipmakers, is expected to underpin growth in 2025 and beyond. 'We believe this year will be better than last, and the years to come will also benefit from the trend of higher equipment spending,' he said. While research and development remains anchored in Singapore, Loh said Malaysia will continue to absorb most of the group's manufacturing investments, with Penang serving as its volume hub for global customers. Deputy Finance Minister Lim Hui Ying, who officiated at the ceremony, hailed the cross‑border listing as a milestone for Malaysia's capital market and a signal of growing economic integration between Malaysia and Singapore. She said UMS Integration's move underscored confidence in the robustness of Bursa Malaysia's infrastructure, investor depth and governance standards, while aligning with the government's ambitions to position the country as a hub for regional and international listings. Lim highlighted Penang's pivotal role in UMS Integration's growth story, noting how the company's expansion has paralleled the state's rise as a global semiconductor and advanced manufacturing hub. 'UMS Integration's continued investment has not only enhanced Penang's manufacturing ecosystem but also created high‑value jobs and opportunities, aligned with the Penang 2030 vision to attract quality, future‑forward investment,' she said. She added that Malaysia's regulatory ecosystem is evolving to support innovation, digitalisation and sustainable finance, and expressed hope that UMS Integration's secondary listing would inspire other regional players with Malaysian ties to consider a presence on Bursa Malaysia. 'This milestone reflects our broader aspiration to connect Asean businesses with global investors, while driving high‑tech and capital‑intensive sectors critical to the region's growth,' Lim said.
