Deep Learning excels in analyzing pictures & videos and creating facsimiles or combining styles. People are using generative AI tools like ChatGPT or Midjourney increasingly frequently. And there is an explosion of simple tools (like the Deep Dream Generator or DeepAI) that use Convolutional Neural Networks to combine your photo with an art style (if you want to do it on your phone, check out Prisma). Here are some example photos.
via SubSubRoutine
The same foundation that allows us to create these cool art amalgamations also can create deepfakes. A Deepfake is precisely what it sounds like ... they use "Deep Learning" to "Fake" a recording. For example, a machine learning technique called a Generative Adversarial Network can be used to superimpose images onto a source video. That is how they made this fun (and disturbing) Deepfake of Jennifer Lawrence and Steve Buscemi.
Another interesting technology can create AI-powered replicas of someone that don't just look and sound like them – they can respond like them too. Examples of this are seen in tools like Replica Studios or Replika. One of the artistic uses people have been exploring recently is getting unlikely characters to sing famous songs. These chatbots have also been used by lonely men and women to create virtual paramours.
The three basic uses of deep learning (described above) are being combined to create a lot of real mainstream applications ... and the potential to create convincing fakes.
Deepfakes can be fun and funny ... but they also create real concerns. They're frequently used for more "nefarious" purposes (e.g., to create fake celebrity or revenge porn and to make important figures say things they never said). You've likely seen videos of Trump or Biden created with this technology. But it is easy to imagine someone faking evidence used at trial, trying to influence business transactions, or using this to support or slander causes in the media.
As fakes get better and easier to produce, they will likely be used more often.
On a more functional note, you can use these technologies to create convincing replicas of yourself. You could use that replica to record videos, send voicemails, or participate in virtual meetings for you. While I don't encourage you to use it without telling people you are, even just using the technology puts you a step ahead.
A Brief Look At Quantum Computing
I am not an expert on quantum computing ... but I saw an impressive photo of Google's new quantum computer, and thought it was worth diving a bit deeper.
Google's computer stands at the forefront of computing technologies. This extraordinary device boasts 70 qubits, a significant improvement over the previous 2019 model, which had 53 qubits. A qubit is the quantum world's answer to classical bits. Not to dive too deep, but as you increase the number of qubits in a model, the possible states a quantum computer can hold simultaneously grows exponentially (due to quantum entanglement,) allowing it to perform faster calculations.
So, while 70 qubits don't sound like that much, it calculates exponentially faster than normal computers. For some context, Google's team used a synthetic benchmark called random circuit sampling to test the system's speed, and the results showed that they could perform calculations in seconds that would take the world's most powerful supercomputer, Frontier, 47 years.
Four years ago, Google announced that they'd reached quantum supremacy, a benchmark demonstrating that a programmable quantum device could solve a problem impossible for classical computers to solve within a practical timeframe. It took less than five years to successfully establish the technological feasibility of quantum computers.
But, it's an exciting horizon for us to walk towards.
Onwards!
Posted at 05:58 PM in Business, Current Affairs, Gadgets, Ideas, Market Commentary, Science, Trading, Trading Tools, Web/Tech | Permalink | Comments (0)
Reblog (0)