Summary of This Week in AI - 30 August 2024

00:00:00 - 00:25:00

In the August 30, 2024, episode of "This Week in AI," hosts Steve Hargadon and Reed Hepler discussed recent advancements in artificial intelligence technology. Google made headlines with new features for Gmail, including a chat bot for email searching and summarizing, as well as the rollout of Google Gems. However, OpenAI faced criticism for delayed rollouts, particularly the voice upgrade, and the departure of significant personnel. The hosts also touched upon the limitations of large language models and the importance of understanding how AI is trained. Steve Hargadon shared his experience using OpenAI and expressed concerns about unrealistic expectations for AI's ability to replicate human intelligence. They discussed various AI tools, including chatbots, and the economic models of content platforms. The effectiveness of first-generation robots and the surge in popularity of character AI were also discussed, with a focus on the importance of human agency in the use of AI tools. The hosts emphasized the role of humans in relation to advanced AI technologies and looked forward to exploring the implications of these developments further.

• 00:00:00 In this section of the "This Week in AI" YouTube video from August 30, 2024, hosts Steve Hargadon and Reed Hepler discuss recent developments in artificial intelligence technology. Google was a major focus of the news, with the announcement of new features for Gmail, including a chat bot for searching and summarizing emails, and the rollout of Google Gems, which functions as a customized version of GPts. Google's improvements to their AI technology, such as the correction of anachronistic depictions of individuals, were also highlighted. OpenAI, on the other hand, faced criticism for delayed rollouts, particularly the voice upgrade, which had been promised for months. The hosts also touched upon the departure of significant personnel from OpenAI and the resulting communication gap.
• 00:05:00 In this section of the "This Week in AI" YouTube video from August 30, 2024, hosts Steve Hargadon and Reed Hepler discuss the latest advancements in artificial intelligence (AI), specifically the new version of Chat GBT 5, which is rumored to have more logic and independent thinking capabilities. Steve expresses his positive experience collaborating with the improved version and shares speculation about the potential introduction of groundbreaking AI technology, possibly even something "dramatic," this fall. The conversation then shifts to the topic of large language models and their inability to accurately identify the number of r's in the word "strawberry." This issue, according to Reed, arises because these models work with tokens rather than words, and the misunderstanding has led to debates about the capabilities and limitations of AI. The hosts also touch upon the importance of understanding how AI is trained and the potential consequences of misusing it.
• 00:10:00 In this section, Steve Hargadon shares his experience using OpenAI, expressing his astonishment at the model's ability to understand and assist him, leading him to question its lack of sentience. He also discusses his concerns about the unrealistic expectations we have for AI to replicate human intelligence, as humans are largely influenced by non-rational factors. Anthropic, another AI company, made news this week by making their cloud artifacts free and transparent, allowing users to see the conversation between the user and the AI model while creating websites or books. Additionally, Anthropic and OpenAI have expressed support for the California AI legislation, with both companies publishing their instructions given to their large language models. Steve Hargadon is curious about the extent to which these models are being shaped to conform to social and political boundaries and looks forward to learning more about the human feedback and instructions that influence their development.
• 00:15:00 In this section of the "This Week in AI" YouTube video from August 30, 2024, Steve Hargadon and Reed Hepler discuss their experiences with various AI tools, including chatbots like ChatGPT, Gemini, and Perplexity. Hargadon expresses his satisfaction with these tools' ability to summarize YouTube videos and articles, saving him time and helping him decide whether to watch full videos or not. However, he notes that none of them can access Substack articles, which is a drawback for him. Hepler adds that while he appreciates the transparency of the tools, they are not accessible to AI tools themselves, requiring more manual interactions. The conversation then shifts to a discussion about the economic models of Substack and Medium, with Hargadon expressing his preference for Medium's single fee for access to all paid content. They also mention a new $16,000 robot and reminisce about Honda's Asimo robot from 2000, which was more expensive and less advanced than current AI-enabled robots.
• 00:20:00 In this section of "This Week in AI - 30 August 2024", Reed Hepler and Steve Hargadon discuss the effectiveness of first-generation robots, specifically Tesla's Cybertruck, and the surge in popularity of character AI. Hepler raises concerns about the Cybertruck's durability and wonders if it will be effective, while Hargadon expresses skepticism about its value. They also discuss the educational and emotional uses of character AI, which people prefer over human interaction due to the ease of text-based communication. Hepler emphasizes the importance of humans having agency in the use of AI tools and the need for educators to prepare students for an AI-driven world. Hargadon suggests that teaching about AI could be a valuable educational moment, but the general understanding of AI among educators may not be high enough for consistent effectiveness.
• 00:25:00 In this section of the "This Week in AI" YouTube video from August 30, 2024, hosts Steve Hargadon and Reed Hepler discuss the role of humans in relation to advanced AI technologies. Hargadon shares his experience of conversing with a language model, where he framed himself as the agent, emphasizing the importance of human involvement in the conversation. He expresses skepticism about AI sentience but acknowledges the potential for productivity enhancement through the use of these tools. The conversation then shifts to the question of how individuals will utilize these emerging technologies, with Hargadon suggesting the possibility of being passive recipients or active agents in the process. Hepler agrees, and they both look forward to exploring the implications of these developments in future episodes.