Why Europe Must End Its 30-Year Digital Winter to Ensure Its Long-Run Future

This Week in AI: Optimism, EU Regulations and Lightweight Bots

artificial intelligence

This week in artificial intelligence (AI) news, a study examined different views on AI’s future, while the EU is moving forward with strict AI laws. Meanwhile, smaller and more efficient chatbots, like Inflection’s Pi upgrade, are making AI cheaper and more accessible for businesses, showing a trend toward cost-saving and energy-efficient AI models.

Debate Over AI’s Impact Continues

The conversation around AI is deeply divided between optimists and pessimists. On one hand, there are AI experts deeply concerned about the technology’s potential risks, fearing its unintended consequences. On the other hand, “super forecasters” and some professionals suggest a more measured approach, emphasizing cautious optimism.

A study by the Forecasting Research Institute has highlighted this divide. It revealed that AI experts generally express greater concern over AI’s potential dangers than their super forecaster counterparts, pointing to a complex landscape of opinions surrounding AI’s impact on our future.

Beth Simone Noveck, a professor at Northeastern University, told PYMNTS that AI, at its core, is sophisticated yet non-sentient software designed to process and analyze data on a large scale. Noveck said that the real potential of AI lies not in fear-mongering about its capabilities but in harnessing its power to address critical global challenges such as inequality, climate change and social justice.

This perspective invites a more nuanced discussion about AI: instead of dwelling on dystopian scenarios, the focus should shift toward leveraging AI as a tool for good. 

EU’s AI Laws Ignite Discussion on Hindering Innovation

The EU has passed new laws to rein in AI, but there are rumbles of concern from business. The European Union’s parliament has passed the first-ever comprehensive artificial intelligence (AI) regulations, sparking both praise for its forward-thinking approach and concerns over potential negative impacts on business innovation.

This groundbreaking legislation targets influential, general AI models and high-risk AI systems, imposing strict transparency and compliance with EU copyright laws. It also restricts government use of real-time biometric surveillance in public spaces to situations involving crime prevention, terrorism counteraction, and tracking major offense suspects. Implementing this law may pose challenges for AI developers and users by narrowing their operational freedoms. 

“Well-designed regulations can enhance trust and reliability in AI, essential for its adoption in business,” Timothy E Bates, a University of Michigan professor specializing in AI, commented to PYMNTS. “Yet, the risk is that too strict or inflexible rules could slow innovation and disadvantage businesses, particularly smaller ones with fewer resources to deal with regulatory demands. It’s vital for regulations to balance setting standards with fostering innovation.

Lightweight AI Alternatives to GPT-4 Level the Playing Field

Chatbots are getting smaller, which could save money and energy. Inflection’s recent Pi chatbot upgrade is one recent example of the trend of developing more compact and cost-effective AI models, making the technology more accessible for businesses.

The chatbot has been updated with the new Inflection 2.5 model, which achieves nearly the same effectiveness as OpenAI’s GPT-4 while only requiring 40% of the computational resources for its training.

Inflection 2.5 boasts enhanced coding and mathematics capabilities compared to its previous version, designed to enable natural, empathetic and secure conversations. The upgraded model expands the range of topics Pi users can discuss, demonstrating that smaller large language models (LLMs) can still deliver strong performance efficiently. 

“Smaller LLMs provide users with more control compared to larger language models like ChatGPT or Anthropic’s Claude, making them more appealing in many situations,” Brian Peterson, co-founder and chief technology officer of Dialpad, a cloud-based AI-powered platform, told PYMNTS in an interview. “They can filter through a smaller subset of data, making them faster, more affordable, and, if you have your own data, far more customizable and even more accurate.” 

Pi’s chatbot may be compact, but it delivers a powerful performance and capabilities. Inflection 2.5 achieves more than 94% of GPT-4’s average performance on benchmarks such as massive multitask language understanding, which assesses a model’s language understanding capabilities. This was accomplished using just 40% of the FLOPS required by the OpenAI model. 

Smaller LLMs, also known as small language models (SLMs), typically have between a few hundred million and 10 billion parameters, requiring less energy and computational resources compared to their larger counterparts.

SLMs make advanced AI and high-performance natural language processing tasks more accessible to a wide range of organizations. The costs associated with SLMs are lower due to the use of more affordable graphic processing units and machine-learning operations.

“We are currently witnessing a Cambrian explosion of small and medium-sized language models in the open-source community,” Akshay Sharma, chief AI officer at Lyric, an AI-based payment technology company, told PYMNTS in an interview. 

While GPT-4 and other large models remain popular, both the enterprise and startup sectors are seeing numerous companies release their own SLMs, Sharma said. Examples include Meta’s Llama2 7b, Mistral AI’s 7b, and Microsoft’s Orca-2.

One advantage of smaller LLMs is their efficiency. However, the growing energy consumption of LLMs is raising concerns among experts and environmentalists. As these AI models become more advanced and widely used, the computational power needed to train and deploy them is leading to a substantial increase in electricity use, contributing to the industry’s growing carbon footprint.

Amazon Unveils AI Tool to Automatically Generate Listings

A new AI tool will let Amazon sellers create listings with just a few clicks. On Wednesday (March 13), Amazon introduced the generative AI (GenAI)-powered feature that allows sellers to provide a link to their own website and automatically generate product listings for the Amazon store.

The company said the new feature would “save selling partners time and effort while creating listings that appeal to customers and drive sales.” The tool is currently being rolled out and will be available to U.S. sellers within the next few weeks. 

This new AI feature builds upon existing tools launched last fall. These tools allow sellers to create product listings by providing just a few words or an image of the product. The AI system then generates a product title, description and additional details, optimizing the product page for better search results.

By using these AI-powered tools, sellers no longer need to manually input all the product information, making the process of creating listings much simpler and faster. Amazon’s introduction of these tools shows the company’s ongoing efforts to use advanced technology to support its sellers and improve customer experience on its platform.