What This Week’s Big AI Announcements from OpenAI and Google Mean for You
Well. This was a big week in AI with both OpenAI and Google making huge announcements. To decipher it all and better understand what truly matters from these big news events, we turned to OpenWater member and AI expert Daeil Kim.
Following is a Q&A with Daeil we conducted over email. Read it see what changes are coming to AI (and your work).
Whoa! It’s a big week in AI. Chat-GPT, Google I/O. What do you think are the biggest bits of news from these two competing AI Platforms?
For ChatGPT, it’s certainly their release of a low latency multi-modal first model with improved reasoning that feels very much like the conversational agent we saw in the movie “Her”, starring Joaquin Phoenix. Their demo was also refreshing in that it was in real-time and not highly choreographed, showing that this future exists today. For Gemini, they also went heavy with the multi-modal capabilities, to some degree recognizing they were in a place where they’re playing catchup to folks like OpenAI, especially in their announcement of Veo (their version of OpenAI’s Sora). The most interesting difference for Google is definitely the array of products that they can integrate their AI into. This is one area where OpenAI will have a hard time competing and this might be why there are a lot of rumors right now of OpenAI talking to Apple to get access to a large product platform.
What are some practical things people can now do with these new features?
For myself, I now have a 24 hour on demand tutor who can teach me anything and does a much better job at understanding my question because it can also understand what I’m looking at. Furthermore, I’m able to just talk to my tutor so I can be coding or writing up a doc while simultaneously asking questions to give me insight into what I’m working on. There’s really no limit now to what these systems are capable of doing so I recommend that we all just roll up our sleeves and start using them anywhere and everywhere.
What are some im-practical things people can do with these features (things that might not occur to them, not necessarily useful to their everyday lives, but nonetheless cool)?
I think that folks who might originally be intimidated by certain tasks can now accomplish them more effectively than before. For example, maybe there’s an amazing website that you love that you might want to emulate. You could now have the camera pointed at the website and your code and it’ll give you detailed instructions while you’re writing to emulate the website that is also within its field of view. We shouldn’t be shy about how we can push the limits of these systems to help us and I would encourage us to try being bold with our requests and you might be pleasantly surprised.
How do these announcements make you (re)consider the current state of AI?
What’s clear has been the focus on multi-modality and how these models are now becoming smarter by integrating across text, video, and audio. The speed with which this has occurred is truly amazing. Moving forward, the big problems that remain will be driven by arguably the next step of AI evolution which will be tied to spatial intelligence. How do we get these AIs in the physical world to learn from that type of data to then accomplish almost any task that a human might be capable of, ushering in the age of advanced humanoid robots. A wonderful TED talk on this is worth watching for some inspiration:
Anything else?
The breakneck pace of technological advancement can be overwhelming, particularly in the realm of AI. However, it’s important to acknowledge that these systems will fundamentally reshape our economic landscape within our lifetimes, on both global and individual scales. Today’s leaders, with their influence and capital allocation, will play a pivotal role in this transformation. Thus, I think it’s imperative to become knowledgeable and proficient users of these systems sooner rather than later, ensuring that our opinions are rooted in firsthand experience.
