The legal discovery process in Google’s antitrust case with the Department of Justice provided a fascinating look into the future of ChatGPT.
A internal OpenAI The strategy document entitled “ChatGPT H1 2025” describes the company’s aspiration to create an “AI super-assistant that deeply understands you, and is your interface with the internet.” It reveals that OpenAI wants ChatGPT to become much more than just a chatbot.
In the first half next year, we will start evolving ChatGPT to a super-assistant. One that knows you and understands what matters to you. It will help with any task a smart, trusting, emotionally intelligent computer could perform. The timing is right. Models such as 02 and 03 can now reliably perform agentic actions, and tools like computer usage can boost ChatGPT’s ability to take action. Interaction paradigms such as multimodality and the generative UI allows both ChatGPT users and ChatGPT to express themselves the best way possible for the task.
This document describes a “super-assistant” as an intelligent entity with “T-shaped skills” that is able to handle both widely applicable tasks and niche tasks. The document says that the “broad part” is about making life easier. It includes answering questions, finding a house, contacting an attorney, joining a fitness center, planning vacations and buying gifts. It also mentions coding, which is a niche task.
It’s clear from the redactions that OpenAI views hardware as essential to their future and that they want people to see ChatGPT as more than just a tool. This tracks with Sam Altman. Recently, a young person said
that ChatGPT was used by them as a “life advisor.” Another part is also included in the strategy document. “Our vision for ChatGPT, however, is to help you in all aspects of your life no matter where you may be. It should be able to answer questions, play songs, and give recipes at home. It should help you find your way, find the best restaurants or catch up with friends when you are on the move. It should help you prepare for a big presentation or take notes during meetings. On solo walks, you can use it to reflect and wind down.
OpenAI is in a shaky position. Altman’s focus is on building data centres because its infrastructure can’t handle ChatGPT usage. In a section describing AI chatbots, the company writes “we are leading in this area, but we cannot rest” and that “growth will not line up for ever.”
The document acknowledges “powerful incumbents” who will leverage their distribution and their own products to their advantage. OpenAI will advocate regulation that will require other platforms to let people set ChatGPT to be the default assistant. Apple is rumored soon to allow iOS users to select Google’s Gemini when they ask Siri questions. Meta AI has also just reached one billion users, largely due to its many hooks on Instagram, WhatsApp, Facebook, and other social media platforms.
The OpenAI document states: “We have everything we need to succeed: one of fastest-growing products ever, a category-defining product, a research leader (reasoning multimodal), a computing lead, a team of world-class researchers, and a growing number of people with agency, who are motivated to ship.” “We don’t depend on ads. This gives us flexibility in what we build. Our culture values self-disruption, speed, and bold moves. “Maintaining these advantages takes a lot of work, but if we succeed, they will last a long time.”
Apple flies the coop: This is the first time since a decade that Apple’s executives won’t be participating in John Gruber annual post-WWDC live podcast. Gruber wrote the viral Apple circles were abuzz with the essay “something is wrong in Cupertino.” It’s not hard to see that he hasn’t publicly linked the critical piece with the company pulling out of his podcast. It says a great deal about Apple’s state when its leaders refuse to participate in a forum that has been friendly for many years.
Overheard
The way we do ranking Sundar explained toDecoder why the company won’t change its search results for President Trump or anyone.
Compared to previous technological changes, I’m more concerned about the labor impact…Yes, people will adjust, but they may adapt too slowly.” – Anthropic Chief Executive Dario Ammodai On CNN, raising alarm about the technology that he is developing.
Meta is a different company from nine years ago, when they fired me. – Anduril’s founder Palmer Luckey. Telling AshleeVance that he will be working with Mark Zuckerberg to produce headsets for the military.
Personnel Log
- The flattening out of Meta’s AI group has taken place, with VP Ahmad Al-Dahle not longer overseeing the whole group. He now co-leads the “AGI Foundations”along with Amir Frenkel, VP engineering, and Connor Hayes, who runs all AI products. All three men report to Meta CPO Chris Cox (19459030) who has taken over the role of Meta CPO. J Allard, Xbox co-founder (19459030), is leading the new ‘breakthrough devices’ group at Amazon. It’s called ZeroOne. According to job listings, one of the devices is smart home related.
- C.J. Mahoney (19459031), a former Trump Administration official. Microsoft has promoted Lisa Monaco to global policy director. She was previously in the Biden administration. Reed Hastings (19659025) He joins Anthropic because he believes in their approach to AI and wants to help humanity advance. (He is joining the corporate board and not the supervising trust board that can hire and dismiss corporate directors.
- Sebastian Barrios previously SVP at Mercado Libre, Chris Rogers, who is currently the chief business officer at Instacart, will replace Fidji Simo as SVP of Engineering. He will be responsible for several areas including ads, game discoveries, and virtual currency. Fidji’ssuccessor at Instacart is Chief Business Officer Chris Rogers, who will replace him. She will officially join OpenAI on August 15th and become its next CEO.
List of links
Click on more:
I always welcome your feedback. Especially if you have any thoughts or story ideas to share. You can reply here or send me a secure Signal message.
Thank you for subscribing.
