TIME outlines three differences of modern artificial intelligence
Modern artificial intelligence differs from previous generations through the versatility of its models, interaction in natural language, and the ability to carry out a sequence of actions, TIME reports. At the same time, such systems still require human judgment, particularly in tasks where understanding context is important.
From narrow systems to general-purpose models
Artificial intelligence has been developing since at least 1956, when a group of computer scientists gathered at Dartmouth College to discuss whether machines could think. Previously, AI mostly operated unnoticed: for example, it predicted customer churn or detected fraud. Such models were usually trained for one narrow task.
Today’s foundation models are large neural networks trained on significant volumes of data and can adapt to different contexts. The same system can help a developer write code, a marketer prepare texts, and an HR specialist create a job description.
More current news is available on the UA.News Telegram channel Telegram.
Language and autonomous actions
Another change has been the ability to interact with AI systems in ordinary language, without programming or rigid menus. Generative AI can create new content, including texts, images, and code. At the same time, because such tools are accessible, companies need to determine when the system’s results can be trusted and when they should be checked or corrected.
The publication also draws attention to agentic AI, which can plan steps to achieve a goal, use the necessary tools, check its own work, and continue carrying out a task. Such an agent can draft an email, open a request, schedule a delivery, and record the transaction without a person clicking the send button.
At the same time, AI capabilities are uneven. A system can produce a complex legal memorandum but misinterpret a contract’s indemnification provision. It can also write working code but fail to account for an implicit requirement that a junior engineer would notice. Harvard researcher Fabrizio Dell’Acqua and his colleagues described this as AI’s jagged technological frontier.
