The wildest week in AI history
Week 20 signalled head-to-head competition between the two biggest players in AI right now. A tectonic shift in search was introduced, and China is showcasing new ways to talk to your loved ones that passed away. I have once again collected and curated the most important stories about AI from the last week.
GPT-4o
OpenAI teased Google by going live with their latest LLM called 4o, where “o” stands for omni. What’s new about GPT-4o is that it’s trained multimodally from the ground up, meaning it understands speech, images and probably video directly without having to translate back and forth to text.
This creates a range of new possibilities and a solution you can talk to without delays. That’s why many have called it a “Here” moment, in reference to the film Here. If you haven’t seen it yet, it’s available on Netflix, but only until 31 May!
Or read this article from the New York Times: https://www.nytimes.com/2024/05/14/technology/ai-chatgpt-her-movie.html
They’re also introducing a desktop app, which I actually already have access to. This means you can now have AI running in the background as a helper you can constantly talk to and ask questions to while you work. Sam Altman believes that in just 14 days it has changed the way he works, as he explains in this clip: https://x.com/tsarnick/status/1790496768851116473
To watch the full interview with Altman from The Logan Bartlett show on Monday, the day the GPT-4o was launched, watch it here: https://www.youtube.com/watch?v=fMtbrKhXMWc
The model will be free and available to everyone! However, it will be launched to paying users first and some of the new features will only be available to paying users, so don’t cancel your subscription!
One of the best demos they did was with the Danish company: Be My Eyes where their upcoming solution with GPT-4o helps blind people understand their surroundings in real time. Watch the demo here: https://www.youtube.com/watch?v=KwNUJ69RbwY
In general, the easiest way to get a good impression of the new possibilities is to watch the long series of demo videos that OpenAI has posted on their site: https://openai.com/index/hello-gpt-4o/
Or check out their Vimeo account, where you can find them all: https://vimeo.com/openai
As the new features are released, I’ll be back with more news.
Google launches AI overviews and Project Astra
Over 2 long hours on Tuesday, Google launched a whole bunch of tech news and what they all had in common was that they were all about AI.
In fact, AI was mentioned no less than 120 times, which is once a minute! If you don’t believe us, check out this clip of them all: https://www.cnet.com/videos/how-many-times-did-google-say-ai-at-io-2024/
Most interestingly, they are now taking the first step to fundamentally change their search engine with the introduction of what they call AI overviews. This has the potential to change how online businesses drive traffic to their websites in the future.
But there were so many things, and most will only become available over the next several months.
Here is a list of 100 different launches from the event: https://blog.google/technology/ai/google-io-2024-100-announcements/
The market for deadbots is exploding
As the technology breakthroughs we’ve seen this week become freely available, it will be even easier to make a digital replica of yourself or others.
And it could even be a deceased son or grandmother. There’s already a big market for this in China, where there is a tradition of “talking to the dead”. Check out this video of a father and his digital bot of his now dead son: https://www.youtube.com/watch?v=SHjKdu8ubh0&t=22s
While it used to cost thousands of dollars to recreate a digital version of a loved one who has passed away, new AI technology has dramatically reduced the price of these types of grief-bots, ghost-bots or dead-bots.
But is this good or bad? And will it become a major phenomenon in our part of the world?
The ethical dilemmas are queueing up. And there’s a lot being written about it this week: https://futurism.com/the-byte/cheap-ai-clones-dead-relatives
MIT has also released a fresh article on the topic: https://www.technologyreview.com/2024/05/07/1092116/deepfakes-dead-chinese-business-grief/
At the University of Cambridge, researchers are looking at the need for security protocols in this area. They have just published this article focusing on the ethical challenges and potential problems these solutions could cause: https://www.cam.ac.uk/research/news/call-for-safeguards-to-prevent-unwanted-hauntings-by-ai-chatbots-of-dead-loved-ones
What are your thoughts?
Wouldn’t it be nice to have a digital avatar of your deceased mum or dad that you could ask for advice? Or is that too spooky and should we leave the dead alone? In any case, it’s a very exciting topic that has never been more relevant than now.
Claude 3.0 Opus finally available in Europe
Since Anthropic launched their 3.0 model in the US, testimonials from users have pretty clearly indicated that this is currently the best model in the world.
We just haven’t been able to test it in Europe until this week. On top of all the other launches, we suddenly got full access even here in Denmark — both via web and via their iOS App.
According to Claude, their model is trained to be more honest, ethical and transparent than ChatGPT. Many also believe it’s better for summaries. And it’s free for now.
Test it yourself and let us know what you think: https://www.anthropic.com/news/claude-europe
That´s all for this hectic week. Please let me know waht you think about the news and the dilemmas that they create, and give me a clap if you like my weekly news.
