AI ARTICLE
What’s happening in the AI World ?
Hi and Happy Sunday! I wanted to catch you up on some of the latest happenings in AI.
Grab some coffee and let’s dive in!
CES 2024 brought a tidal wave of tech announcements last week.
But the AI news has kept pouring in even after the event ended.
This past week alone was jam-packed with so many exciting developments across research, products, policy…you name it!
I’ll start with some of the flashy new AI consumer products and work my way into the intriguing research breakthroughs.
Samsung’s AI-Packed Galaxy S24
As you probably heard, Samsung unveiled their latest flagship phone — the Galaxy S24.
It’s brimming with ambitious AI features making it clear that Samsung is charging full steam ahead to position themselves as THE AI phone company.
While Apple has been surprisingly slow to talk about AI in their products, Samsung is wanting the S24 to be known as the phone that helps you “do more with less effort” thanks to AI.
And let me tell you all, some of the new capabilities look pretttty slick!
Real-Time Translation
One feature I’m super excited about is real-time translation right on your phone.
Whether you’re chatting via text messages and social media or having an old-fashioned phone call, the S24 will automatically translate between languages in real-time!
So if you’re talking to your Spanish-speaking abuela, the phone will instantly translate back and forth between you speaking English and her speaking Spanish. How cool is that?
I don’t know about you, but language barriers have always made people feel awkward. But no more thanks to AI!
The translations work for up to 13 languages at launch and the processing happens directly on the device so you don’t even need an internet connection.
Built-In AI Chip
This is possible because the Galaxy S24 actually has its own dedicated AI processor called the Gemini Nano chip.
It handles real-time translation and other features right on the device itself, no cloud required.
Some other handy features powered by the on-device AI chip include:
- Summarizing group text messages read aloud while you drive so you don’t get FOMO - Transcribing voice memos and summarizing the key details so you don’t have to listen to long recordings - Suggesting photo edits to fix things like glare and reflections with just a tap
When you want to tap into more advanced AI requiring extra computational horsepower, the phone can connect to the cloud. For example:
- Circling something in a photo and asking a question to magically get info from the web based on what you highlighted - Adding artificial slo-mo to videos by using AI to generate extra frames - Moving objects around in photos and having AI generate realistic backgrounds in the empty areas
I’m not gonna lie, I think the image editing features could lead to some wild and creative photography!
Killer AI-Enhanced Camera
Of course no Samsung phone would be complete without a KILLER camera
The S24 Ultra model has a 200MP main camera powered by what they call a “Pro Visual AI engine.”
Besides helping you look fab in selfies, the AI-enhanced processing means your photography skills don’t have to be pro level to end up with shots that look pro.
We’re talking automatically enhancing portraits, suggestions for artistic edits, adding depth and focus changes…all with just a couple taps.
The AI also lets you do things like smoothly add slo-mo to videos without needing fancy high FPS(Frame Per Second) camera gear.
It analyzes each frame, uses AI to generate extra frames, and boom — cinematic Matrix-style slo-mo at normal 30fps.
There’s also new AI video and photo “generative” features where you can cut someone out of a pic and the AI will realistically fill in the background. Or you can duplicate objects and people.
Availability & Pricing
Pre-orders kick off Jan 17 with delivery on Jan 31. Pricing ranges from $999 to $1,299 depending on the model.
There is one annoying caveat though. After 2 years, you’ll have to pay extra to keep access to the AI features that require cloud connectivity.
Sneaky way to get you hooked before making you pay, Samsung!
OpenAI’s Sam Altman on the Future of AI Models
Whew! That was a lot just on Samsung’s new phone. But sooo much more has been going on!
OpenAI CEO Sam Altman did a bit of a media tour last week chatting about AI safety and capabilities with Axios, Bloomberg, Bill Gates, the World Economic Forum, and more.
I combed through all his latest interviews to pull out the highlights for you!
Customized AI Assistants
One big theme was AI models evolving to become more personalized for each user:
“ChatGPT will have to evolve in uncomfortable ways sometimes based on people’s different value systems or people’s different cultures or people’s different countries.”
So in the near future, AI chatbots like ChatGPT won’t be a one-size-fits all model trained on the entire internet.
Instead, the assistant you chat with will be tailored to your specific location, culture, values, beliefs, etc.
So, that means, my ChatGPT assistant would give different responses than yours!
When asked about what milestones OpenAI is focused on for the next 2 years, Sam said things like reasoning ability, customizability, and integrating external user data are top priorities for improvements.
AGI Won’t Radically Transform Society
I found Sam’s perspectives on artificial general intelligence pretty interesting too.
There’s a camp of people ringing warning bells about AI taking jobs and radically changing society.
But Sam isn’t quite as much of an alarmist:
“AGI will get developed in the reasonably close-ish future and it’ll change the world much less than we all think. It’ll change jobs much less than we all think.”
“I think it’s good that people are scared of AI to some degree because I think that helps us approach it more thoughtfully and more cautiously.”
He seems to believe developing AGI is inevitable in the coming decades.
But also thinks fears about mass job loss and runaway AI are a bit overhyped. The impacts will be real but more nuanced. What do you think?
The Data Question
Lots of debate lately about whether companies like OpenAI should be able to scrape datasets without explicit permission.
Sam clarified that scraping every last site isn’t a priority despite what some critics claim:
“We do not want to train on the New York Times data for example. More generally, we’re getting to a world where it’s been like data, data, data. You just need more, more, more. You’re going to run out of that at some point anyway. So a lot of our research has been, how can we learn more from smaller amounts of very high-quality data?”
Military Applications Allowed
One ethics-related policy shift that raised eyebrows was OpenAI updating their content guidelines to ALLOW some military applications that don’t violate human rights or develop autonomous weapons:
“Our policy does not allow our tools to be used to harm people, develop weapons, or communication surveillance or to injure others or destroy property. There are however National Security use cases that align with our mission for example, we are already working with DARPA to Spur the creation of new cyber security tools…”
What do you think about green lighting military usage in some contexts like cybersecurity?
Tricky debate with reasonable arguments on both sides in my opinion.
Microsoft Upgrades CoPilot with GPT-4
Pivoting to Microsoft news now…
Microsoft quietly upgraded its CoPilot coding assistant web app to leverage OpenAI’s latest GPT-4 model!
For context, GPT-4 access normally requires a $20 per month ChatGPT Plus subscription.
But Microsoft is offering up the GPT-4 goodness entirely free in CoPilot.
Though with limited functionality I’d assume. Still though — cheers to free access to what’s normally paywalled AI power!
They also launched something called Copilot Pro which includes extra goodies like GPT fine-tuning tools and AI helper features across Word, Excel, Outlook and PowerPoint.
But that will run you $20 per month.
Facebook…I mean Meta — Charges Into LLaMA
Meta has fully embraced AI as the next major platform shift.
Mark Zuckerberg posted on Instagram that they are hard at work on their internal LLaMA (Large Language Model Meta AI) initiative:
“It’s become clearer that the next generation of services requires building full general intelligence — building the best AI assistant, AIs for creators, AIs for businesses, and more. This technology is so important and the opportunities are so great that we should open source and make it as widely available as we responsibly can so everyone can benefit.”
They are building a massive compute cluster to train models even larger than GPT-4, planning for 600,000 GPUs by end of this year!
Zuck did emphasize intentions to open source the work so others can build on it.
Though “responsibly” is a squishy term when it comes to AI safety considerations. Fingers crossed they implement sufficient guardrails!
Other AI Product Launches
We've covered a TON of ground already. Let’s speed round through some other new AI product launches in recent weeks:
- Amazon added a shopper question answering feature to their mobile app powered by product review analysis. LINK
- Adobe is working on AI audio clean up features in Premiere Pro like dialogue noise reduction. LINK
- Video platform Runway launched a wild new Multi-Motion brush for animating different objects in a scene independently.
The Bottom Line
If you made it this far, hopefully you feel well equipped with both knowledgeable and nuanced perspectives on the latest happenings in AI. Let me know which topics grabbed your attention most! I’m always down for a riveting discussion. As always, thank you for reading. Cheers!!!
Visit us at DataDrivenInvestor.com
Subscribe to DDIntel here.
Have a unique story to share? Submit to DDIntel here.
Join our creator ecosystem here.
DDIntel captures the more notable pieces from our main site and our popular DDI Medium publication. Check us out for more insightful work from our community.
DDI Official Telegram Channel: https://t.me/+tafUp6ecEys4YjQ1