Weeks Count:
0 weeks: This current week.
1 week: The next week following the current week.
N weeks: N weeks from the current week.
OpenAI: Refers to the company OpenAI.
New Voice Mode:
The version demoed during the 4o keynote.
The specific name (4o, 4.5, Omni, etc.) does not affect whether it qualifies; it must simply embody the core functionalities as demonstrated.
Allowances for minor differences from the demo, such as slightly higher latency or less expressiveness.
Must allow for interruption mid-sentence to qualify as the "new voice mode."
Broader Audience:
Defined as random ChatGPT users who are not affiliated with OpenAI through personal or professional connections, including employees, partner companies, friends, or family.
Even a limited rollout counts if at least one non-affiliated user gets access to the alpha.
Confirmation of rollout to broader audience needs to be verifiable through a video demonstration by a user or widespread reports from multiple sources confirming access.
1,000
1.00You can now vote on when a significant portion of ChatGPT plus users will gain access to the new voice mode (so no alpha, closed beta, and so on, we talking actual rollout)
https://manifold.markets/Flowers/when-will-the-new-4o-voice-mode-fin
https://x.com/wongmjane/status/1805873170957635795?s=46
This person posted a video on their x account how they have access to the alpha. OpenAI does not have an official partnership with Meta, the company she is working for.
Even a limited rollout counts if it includes access for at least one non-affiliated user.
So it does counts and this week resolves as yes. But we will continue this market until a "broader audience" gets access. For that reason, above rule now gets deleted and we are specifically only counting a broader audience defined by the other rules.
@Flowers ah, that's not exactly possible if I'm understanding you correctly. If you resolve that option to yes, the entire market ends. You would need to make a new market for the broader release.
@Joshua Oh I see. But then anyways, as per the original rules this should be resolved altogether as yes, right? What do you reckon?
Sorry, but nothing ever happens:
We had planned to start rolling this out in alpha to a small group of ChatGPT Plus users in late June, but need one more month to reach our bar to launch. [...] As part of our iterative deployment strategy, we'll start the alpha with a small group of users to gather feedback and expand based on what we learn. We are planning for all Plus users to have access in the fall. Exact timelines depend on meeting our high safety and reliability bar.
So they're saying late July-early August at best for a public alpha test, and openly floating the possibility of further schedule slippage if they fail to meet their "high safety and reliability bar".
However, alpha access might also count:
Even a limited rollout counts if it includes access for at least one non-affiliated user.
Incredible that I believed them and still, by sheer luck, made a profit.
I guess this whole affair roughly cancels out my truly terrible luck betting on the GPT-4o release (I manged to bet the exact opposite of how each market resolved). Still, a bit unnerving. My Bayes score has definitely taken a hit.
@Flowers If the feature is not by 18 weeks, does it resolve to 18 weeks? or to N/A? When it resolves it has to sum to 100% with this market mechanism
Might be worth adding a flat "November or later" option.
It'd also be handy if the weeks were labelled with absolute dates. Having them relative to the market's creation adds a fair bit of mental overhead in calculating stuff, and that'll only increase as time goes on. [Edit: there's also potential ambiguity in e.g. which day weeks start on, right now. What day does Week 0 end?]