Welcome to our weekly roundup of premium-grade AI news.
This week we wondered if AI is the reason why we can’t find any aliens.
Google’s robot will beat you at table tennis.
And Gemini gets a voice while OpenAI stutters.
Let’s dig in.
Is anyone out there?
There are billions of stars and statistically a fair amount of habitable planets, so why haven’t we detected signs of alien civilizations?
Is there some extinction event that inevitably wipes out a civilization before it can head out to the stars? A new thesis for the Fermi Paradox adds a twist to these questions and asks: Is AI a Great Filter or a cosmic colonizer?
But is it reasonable to assume that AI is as dangerous as some claim? A new study challenges the narrative of AI posing an ‘existential threat’. So maybe we can chill now and stop worrying about Skynet becoming a reality.
“ChatGPT, generate a meme from 2030″ https://t.co/Y0WecFOykg pic.twitter.com/gFK7PYXGPc
— AI Notkilleveryoneism Memes ⏸️ (@AISafetyMemes) May 30, 2024
Anything you can do…
The engineers at Google’s DeepMind insist they’re doing serious research, but their latest project must have been a lot of fun.
DeepMind built a table tennis robot and it’s pretty good. It beats newbie players 100% of the time and the demo videos show it giving its trainers a decent challenge too.
If an AI model had free reign to research whatever it wanted to, what could it discover? Sakana AI developed The AI Scientist, a framework that uses AI models to perform fully automatic scientific discovery.
The AI Scientist isn’t perfect, but it generates academic-quality research papers with novel discoveries at $25 a pop. Imagine if we could put a super smart AI in a lab and tell it to research a cure for cancer.
OpenAI voices its concerns
For weeks OpenAI has told us that we can’t get to play with ChatGPT’s voice assistant yet because it isn’t ‘safe’. This week we got to see some of the weird voice assistant glitches that OpenAI red teamers detected during testing.
Besides a tendency to copy your voice and randomly shout things, there are other issues they will have to fix before we finally get our chatty ChatGPT.
For now, I’ll keep updating the ChatGPT app every few minutes until the voice feature appears.
Gemini speaks!
At the Made By Google launch event, Google showcased new AI assistant features that are coming to Android phones. It also did a demo of Gemini Live, Gemini’s voice assistant.
It has 10 voices to choose from and, unlike OpenAI’s version, it’s rolling out to Google Advanced subscribers as we speak.
True to form, OpenAI found a way to dampen the mood of Google fans. Its updated ChatGPT-4o-latest model just knocked Gemini 1.5 Pro off the top spot on the LMSYS leaderboard.
Bad AI
The dual-use nature of AI came into stark focus this week as a dark web study exposed a surge in AI-generated child abuse content. A UK man is facing charges that challenge the ability of current laws to keep up with new digital crimes.
Why are AI image generators so good at making abhorrent images of children? It turns out that the makers of the models inadvertently trained them to do it. With the genie out of the bottle, it’s a problem that may not be fixable.
AI Summits
Here are some exciting AI events that are happening soon.
- AI and Big Data Expo Europe: Gain valuable insights into the latest Artificial Intelligence and Big Data trends and how they are shaping the future of business.
- DigIT Pharma & eHealth 2024 Conference in Dusseldorf: Explore the latest developments in cutting-edge healthcare technology, including generative AI.
- Marketing AI Conference (MAICON) 2024 in Cleveland, Ohio: Explore the exciting world of AI-powered marketing strategies and tools.
- AIAI Berlin 2024: Explore the latest developments in AI across multiple domains and industries.
In other news…
- Palantir and Microsoft partner to deliver enhanced analytics and AI services to classified networks for critical national security operations.
- Trump falsely claims a crowd photo from Harris’ campaign rally in Detroit was created using AI.
- Huawei’s Ascend 910 launches this October to challenge Nvidia’s H100.
- Deep-Live-Cam goes viral, allowing anyone to become a digital doppelganger.
- AI predicts 3D structures of receptors for faster drug development.
- Elon Musk’s xAI released Grok 2. It smashes benchmarks and its powerful image generation doesn’t seem to have much of a filter.
wow, grok 2 will literally make any image pic.twitter.com/gkaJTsMJuI
— Matteo Pellegrini ⚡️ (@matteopelleg) August 14, 2024
And that’s a wrap.
What’s your take on the distinct lack of alien life? Did their AI take them out? Are we next? Or maybe they took one look at our social media platforms and decided to block our calls.
Thanks to Google we finally have an AI voice assistant. The only problem is, I’m not sure I’m ready to migrate to Android. Let’s hope Apple gives Siri a boost with some new voices.
Have you noticed an improvement in ChatGPT’s performance with the model upgrade? It seems to get some classic problems right that other models struggle with. I’m hoping Sam is just gaslighting us by saying we won’t get GPT-5 this year.
Our conversations continue on X! Follow us for the latest AI insights, cool tech updates, and maybe a meme or two. We’d love to hear your thoughts, so don’t be shy—drop us a message or tag us in your AI musings. See you there!
