“Catastrophic effects”: Can AI decrease human cognition?

Exploring below the surface of AI headlines.

Summaries | Insights | Points of Views

In Today’s Edition

• “Catastrophic effects”: Can AI decrease human cognition

• Nvidia’s Chat with RTX is a promising AI chatbot that runs locally on your PC

• ChatGPT will now remember — and forget — things you tell it to

AI impacts

“Catastrophic effects”: Can AI decrease human cognition?

Summary - In a scientific paper published in the journal Neuropsychology, Professor Umberto León Domínguez from the University of Monterrey explores the potential impact of AI on human cognition. Domínguez raises concerns about the risks posed by AI chatbots to higher-order executive functions. He also discusses the concept of AI as a "cognitive prosthesis" and warns of the dangers of cognitive offloading, where humans become overly reliant on AI, potentially leading to a decline in critical thinking and problem-solving abilities. Domínguez emphasizes the importance of actively engaging in cognitive effort for the effective development of intellectual capabilities.

Buoy points:

• Concerns about cognitive impact: Domínguez discusses the potential negative effects of AI on human cognition, particularly higher order executive functions (reasoning, problem-solving).

• AI as a cognitive prosthesis: AI could potentially perform complex cognitive tasks for humans, acting as a cognitive crutch.

• Cognitive offloading risks: There's a risk that over-reliance on AI could lead to a decline in human cognitive abilities.

• ChatGPT's unique capabilities: Unlike traditional tools, ChatGPT can independently generate ideas and make decisions, which amplifies the potential for cognitive offloading.

• Need for cognitive effort: Domínguez emphasizes the importance of engaging in cognitive effort, especially during formative years, to develop intellectual capabilities.

• Potential for societal impact: If reliance on AI becomes widespread, it could lead to a society with diminished critical thinking and creativity.

POV - Like most human innovations before, AI and chatbots will have its pros and cons. Some could argue that the broader use of advanced calculators starting several decades ago caused cognitive offloading and skill decay, however, look where we are now - human advancement in math and sciences seems to progress faster than ever. I think a greater determinant of AI’s impact is how it is used, and dare I say, “controlled” to be used in society. In other words, the governance of it to be used for human benefit. It’s a delicate balance. What are your thoughts? Is there merit to the study?

AI runs local

Nvidia’s Chat with RTX is a promising AI chatbot that runs locally on your PC

Summary - Nvidia's Chat with RTX is an early demo app that allows users to run a personal AI chatbot on their PC, leveraging the power of an RTX 30- or 40-series GPU with at least 8GB of VRAM. The app enables users to feed it YouTube videos and personal documents to create summaries and obtain relevant answers based on their own local data. While the app shows promise for data research and analysis, it is currently only in its early stages. It does offer us a peek into the potential of local AI chatbots.

Buoy points:

• Local AI Chatbot: Chat with RTX runs locally on a PC, using the power of an RTX 30- or 40-series GPU to process data and generate responses without relying on cloud-based services.

• Retrieval-Augmented Generation (RAG): It uses retrieval-augmented generation (RAG), NVIDIA TensorRT-LLM software, and NVIDIA RTX acceleration to bring generative AI capabilities to local, GeForce-powered Windows PCs.

• Data Analysis and Integration: It supports file formats such as .txt, .pdf, .doc/.docx, and .xml. It also enables integration of information from YouTube videos and playlists for contextual queries.

• System Requirements: Requires GeForce RTX 30 Series GPU or higher with a minimum 8GB of VRAM, Windows 10 or 11, and the latest NVIDIA GPU drivers.

• Technical Limitations: Despite its potential, the app is currently in an early developer demo stage, with known issues and limitations. It requires specific hardware and installation processes, and there are bugs affecting its functionality.

• Try it now: If you have the system requirements met, try it out here on Nvidia’s site.

POV - Local AI will be hugely desired. Many companies and individual users cannot risk exposing their proprietary or personal data to AI that runs elsewhere. Plus, this might take us another step closer to having powerful models on smaller mobile devices, even though it may be a while before we shove the power of RTX in a smartphone. Are you ready to try Chat with RTX?

ChatGPT

ChatGPT will now remember — and forget — things you tell it to

Summary - OpenAI has introduced new "memory" controls for ChatGPT, allowing users to instruct the AI chatbot to remember specific information or forget certain details in future conversations. This feature, currently in testing, enables ChatGPT to provide more relevant responses by retaining information across chats. Users can explicitly ask ChatGPT to remember something or let it learn and pick up details on its own. The memory feature can be useful in various contexts, from personal preferences to business-related tasks. OpenAI acknowledges privacy concerns and provides options to disable or delete memories, as well as a Temporary Chat feature for a privacy-preserving experience.

Buoy points:

• Memory Controls: Allows users to instruct ChatGPT to remember specific information or forget details in future conversations.

• Personalization and Context: Users can personalize conversations by instructing ChatGPT to remember preferences or highlight specific details for tailored responses.

• Business Use Cases: Memory in ChatGPT can be valuable for businesses, such as remembering voice and formatting preferences for blog posts or knowledge of programming languages and frameworks.

• Separate Memories: ChatGPT and GPTs have independent memories that are not shared between them.

• Privacy Considerations: OpenAI addresses privacy concerns by providing options to disable or delete memories and taking steps to avoid proactively remembering sensitive information.

• Temporary Chat Feature: OpenAI introduces the Temporary Chat feature, offering a privacy-preserving experience by starting each conversation with a clean slate and not accessing previous discussions or memories.

The link to OpenAI’s announcement is here.

POV - There have been numerous times when I’ve asked ChatGPT if it still has recall of instructions I’ve given much earlier in a chat session - often, the answer was yes, but the output was ‘iffy’, at best. I’m excited to test out the new memory feature. It will save a lot of re-iteration, especially for long research projects.

Don’t forget to check out other AI headlines in The Ocean.

The Ocean is designed to help you stay informed with the wide array of AI headlines, giving you just the snippet of news that you need if you are short on time, but also gives you one-click ease if you want to dive deeper.