Your choice of AI news resource is impeccable! Welcome to our weekly AI news roundup.
This week in AI we saw the ongoing soap opera drama between Microsoft and OpenAI continue to unfold. We may have to come up with a celebrity couple name for them.
Media companies scrambled to block AI bots and sue the companies behind them. AI has got us worried the kids aren’t all right, and AI is solving medical challenges as fast as it creates psychological ones.
Here’s a wrap of this week’s AI news.
Is it splitsville for Microsoft and OpenAI?
The slow-motion train wreck of the Microsoft OpenAI partnership continued to play out this week. Microsoft is one of the biggest investors in OpenAI but a few days ago it started selling Databricks software on its Azure platform in direct competition with OpenAI’s offering.
OpenAI needs to find a way to actually start making a profit and it’s been heavy going with companies like Meta just handing out their AI models for free. OpenAI finally released its ChatGPT 3.5 Turbo model for fine-tuning and the claimed performance is impressive.
AI is making lawyers a lot of money
It seems like everyone is worried about AI taking their jobs except the lawyers. Legal wrangling over how to handle copyright and AI is giving lawyers plenty of billable hours.
Can an image created by AI be copyrighted? Can you actually own the rights to artwork produced by AI? A US judge made a landmark decision that changes everything.
That decision seemed pretty final, but there’s a lot of uncertainty over how copyright relates to bots scraping data to train AI models. The New York Times may join the long line of people suing OpenAI for scraping its news archive.
If The NYT goes ahead with the lawsuit, OpenAI may have to hand over an eye-watering amount of money. And the money might not even be the biggest consequence of this decision.
I’m not a robot. No, really, I’m not!
Earlier this year a GPT-4 bot hired a worker on MTurk to solve one of those annoying “prove you’re not a robot” CAPTCHA challenges. It turns out that AI doesn’t need humans to do that anymore. New research shows AI outperforming humans in solving CAPTCHAs.
The improved AI performance is interesting, but the stats showing just how bad humans are at proving they’re human were surprising. AI taking over the world is going to be easier than we thought.
It’s not just those workers on MTurk that will see AI affect their job prospects. IBM released a report showing that up to 40% of employees will need to reskill in the next 3 years because of AI.
Think your job is safe because you listened to your guidance counselor and studied something in the STEM field? You might want to read what the report says about that.
Is AI peaking or just getting started?
When even your grandma is explaining AI better than Kamala Harris then you know the AI hype is real. This last year has seen a mad rush to try out ChatGPT with everyone talking about AI. But there are some interesting signs that the hype surrounding generative AI may be starting to settle. Or is it?
Not according to Nvidia’s CEO. The company shot the lights out with its latest Q2 financial report with year-on-year revenue doubling . The speed at which data centers are snapping up Nvidia chips must be making Intel nervous. Nvidia even had to be a little sneaky to sell their chips to China.
Is AI messing our kids up?
Remember how you used to lie to your kids about Santa and the Easter Bunny? Well, AI is a lot better at convincing kids of things that aren’t true. People are freaking out about how little we’re doing to protect them from the potential dangers of AI.
Teachers are having a tough time too. While trying to understand what their Gen Alpha students are actually saying, they also have to work out how to use AI in the classroom. It turns out there’s good news and bad news for teachers and students.
Meta has just released an amazing AI translator that does variations of text-to-speech and speech-to-speech in 100 languages. Now if they could add an option for understanding the ever-evolving slang kids are coming up with then that would be truly revolutionary.
Introducing SeamlessM4T, the first all-in-one, multilingual multimodal translation model.
This single model can perform tasks across speech-to-text, speech-to-speech, text-to-text translation & speech recognition for up to 100 languages depending on the task.
Details ⬇️
— Meta AI (@MetaAI) August 22, 2023
Who would your favorite AI vote for?
Artificial intelligence is just software, right? Just facts and figures coded in ones and zeros, so it shouldn’t be biased. Right? Maybe not. A new study showed that the most popular chat AIs are either left or right of the political spectrum.
Will our future society be left-wing liberal or right-wing authoritarian? It depends on which of these biased models eventually take over. You’d be surprised at who Meta’s model would vote for.
Do you want the good news or the bad news?
AI hasn’t solved our environmental problems yet, but it’s trying. While the data centers are doing their bit to add to global warming, at least AI is getting better at giving us the bad news faster.
Google’s AI flood detection system can predict floods four days in advance and AI is being used to create an accurate tsunami warning system.
It can’t fix those problems yet but at least AI can help us be better prepared.
AI is slaying in the field of medicine
Did I even use that word right? Either way, doctors are using AI to do some awesome things and make a real difference in the lives of patients.
This week we saw AI used to enable near-telepathic communication in ALS sufferers and using AI to give a stroke survivor her voice back.
Doctors found a way to use AI evaluation of eye scans for early detection of Parkinson’s Disease.
AI tech only became available towards the back end of a 10-year brain study but it helped neuroscientists get some interesting insights into the human brain.
We also saw how a new AI tool can tell how your heart age has been affected by drinking, smoking, diet, and your unused gym membership. We don’t think there’s a mad rush to try this one out quite yet.
In other news…
Here are some of the other AI bits and pieces that got us clicking through:
- Meta et al keep saying their AI models are “open source”. Here’s Wired’s take on the myth of open source.
- IBM is using AI to translate COBOL into Java because no one can remember how to program in COBOL anymore.
- Stability AI collaboration with Nvidia sees a boost in Stable Diffusion XL image generation speed.
- YouTube wants to hop on the AI music train but they want to do it “responsibly”.
And that’s a wrap of this week’s AI news. Did we miss a great AI story that you think we should have covered? Let us know.
And next time you’re browsing the internet and have to solve a CAPTCHA, please take your time and put in a little more effort. We need to retain some measure of pride for Team Human.
