AI in February: Smarter, Faster… Still a Bit Clueless?
OpenAI Unveils Deep Research
OpenAI has introduced Deep Research, an AI tool designed to streamline online research and generate concise reports. Launched on February 2, 2025, it can analyze text, images, and PDFs, providing citations for sources. During demonstrations, it showcased its ability to generate comprehensive profiles and scenario-based questions, such as vetting a Senate nominee. Though promising for sectors like finance, law, and science, Deep Research is not without limitations, particularly regarding accuracy and source credibility. Available to ChatGPT Pro subscribers for $200 per month, it marks a major step in AI’s role in autonomous research.
BBC Uncovers AI News Inaccuracies
A BBC study has found significant inaccuracies in AI-generated news summaries from models like ChatGPT-4o, Microsoft Copilot Pro, Google Gemini Standard, and Perplexity. Over 30% of responses contained major errors, including misquotations and outdated information. The study, which examined 100 AI-generated summaries of trending news topics, raises concerns about AI’s reliability in delivering accurate news content. The BBC warns that over-reliance on these tools could mislead audiences, especially when AI models cite trusted sources inaccurately. This growing issue highlights the challenges media organizations face as they integrate AI into their workflows.
The New York Times Embraces AI in Newsroom
Amid concerns over AI-generated inaccuracies, The New York Times is integrating AI into its newsroom workflows with caution. The publication has introduced a tool called Echo to assist with summarizing articles, briefings, and company activity. While AI tools will support tasks like editing and SEO-friendly headline generation, the publication has set strict limitations: AI cannot draft full articles, bypass paywalls, or significantly alter content. The Times is also offering AI training for staff and has approved select AI tools such as GitHub Copilot and Google Vertex AI. These advancements come as the newspaper continues its lawsuit against OpenAI and Microsoft over alleged unauthorized content use, underscoring the tension between AI adoption and journalistic integrity.
Paris AI Summit Highlights Regulatory Divide
As media and tech industries grapple with AI’s role in content creation, world leaders and executives are tackling broader policy challenges at the Paris AI Summit (Feb. 10-11). France has announced a €109 billion ($113 billion) private-sector investment in AI, along with a $400 million public-private partnership with Germany, Google, and Salesforce for open-source AI projects. The summit highlights a regulatory divide: while Europe favors stricter AI regulations, U.S. President Donald Trump is pushing for a more lenient approach to maintain competitiveness. Discussions include concerns about labor displacement and pressure from the U.S. to weaken the EU’s AI Act. Notable attendees include China’s Vice Premier Zhang Guoqing, U.S. Vice President JD Vance, OpenAI’s Sam Altman, and Google’s Sundar Pichai. The outcomes of this summit could shape global AI policy for years to come.
Musk’s Grok 3 Joins AI Race
While governments debate AI regulations, the competition among AI models continues to heat up. Elon Musk’s xAI has launched Grok 3, positioning it as a competitor to OpenAI’s ChatGPT and Google’s Gemini. The model, which includes Grok 3 Reasoning and Grok 3 Mini Reasoning, is designed to enhance problem-solving and reduce hallucinations. Benchmarks show Grok 3 is competitive, with former Tesla AI director Andrej Karpathy stating it is on par with OpenAI’s o1-pro model ($200/month) and slightly better than DeepSeek-R1 and Gemini 2.0 Flash Thinking. However, OpenAI claims its unreleased o3 model outperforms Grok 3 in math and science. Available only to X Premium+ subscribers ($50/month), Grok 3 remains a contender but has yet to surpass its rivals.
