9 months ago
46 ARTICLE AD
Web browsing is not new to ChatGPT. The feature is currently available to ChapGPT Plus and Enterprise users who even have the option to “Browse with Bing”.
ChatGPT creator OpenAI is reportedly working on an autonomous artificial intelligence assistant system with the ability to take control of a device and perform tasks. This is according to a report by The Information citing a source with knowledge of the matter.
The action agent software is expected to automate complex personal and work tasks on behalf of users without constant supervision. It will handle web-based tasks such as gathering public data about companies, creating travel itineraries, and booking flights.
While the Microsoft-backed startup is yet to confirm or deny the news, it seems logical that following the relative success of generative AI models by various tech companies, the next step would be AI assistants.
Web browsing is not new to ChatGPT. The feature is currently available to ChapGPT Plus and Enterprise users who even have the option to “Browse with Bing”. This allows the chatbot to provide more accurate and real-time information.
The development of AI agents with the ability to control users’ devices and access the web could serve as a significant step toward creating fully autonomous digital assistants. Such assistants would be invaluable for both personal and professional use.
For personal use, being able to automate repetitive tasks and access information and services instantly would be highly convenient. On the professional front, being able to automate tedious, repetitive tasks not only reduces the possibility of error but also leaves workers with more time to focus on higher-level tasks.
It is unclear what specific plans OpenAI has for these reported AI agents. The Information’s report asserts that they will be able to operate a user’s device in order to complete a requested task and cites an example in which a user asks the agent to copy data from one platform to another.
Given enough access, the AI agent would be able to replicate physical interactions with a device, such as swiping, tapping, clicking, double-clicking, and typing.
While the potential uses and benefits are irrefutable, there are also privacy and ethical concerns to consider. For some, the idea of surrendering control of their personal device – which contains sensitive and personal information – can be somewhat unappealing.
Available generative AI systems still require connection to massive cloud data centers for some complex operations. Even though some AI functions can run entirely on user devices, an AI action agent such as the one OpenAI is reportedly working on is unlikely to run without cloud access.
At present, the idea of granting a corporate AI system unlimited access to personal information is already a privacy concern. Pair that with internet access and an autonomous AI action agent without proper safeguards, and we have a potentially far-reaching new set of cyber threats waiting to happen.
