#014 Special Feature - X (Twitter) Plans to Use Your Data to Train Its AI Models, What is Algorithm of Thoughts (AoT), Pentagon Using AI to Stay Ahead of China and Russia.
Fresh & Hot curated AI happenings in one snack. Never miss a byte 🍔
This snack byte will take approx 6 minutes to consume.
AI BYTE 1 📢 : How X (Twitter) Plans to Use Your Data to Train Its AI Models?
⭐ Are you a user of X, the social network owned by Elon Musk?
If so, you may want to pay attention to its recently updated privacy policy. It reveals that X plans to use your data and other publicly available information to help train its machine learning and AI models.
X is not the only company that uses public data for AI purposes. Musk has previously accused other tech giants of leveraging Twitter to train their AI models, even for alleged illegal use of Twitter data.
He also blamed unknown entities for scraping Twitter data, which also may have been for the purpose of training artificial intelligence large language models.
But what is X’s AI project exactly?
According to its website, xAI is a separate company from X Corp., but it "will work closely with X (Twitter), Tesla, and other companies to make progress towards our mission. The mission is to create “a general artificial intelligence system that can do anything humans can do, and more.”
Musk has ambitions to enter the AI market and compete with the likes of Google, Facebook, and Amazon. He has also expressed his concerns about the potential dangers of AI and the need for regulation.
He claims that xAI will be “aligned with human values and can be trusted by humans.”
But how much can we trust X with our data? The privacy policy update states that X will use "just public data, no DMs or anything private."
However, it also says that X may share our data with “third parties that help us provide, improve, protect, or promote our Services.” Who are these third parties and what are their intentions?
As users of X, we should be aware of how our data is used and what impact it may have on the development of AI.
We should also demand more transparency and accountability from X and other companies that collect and use our data for AI purposes.
After all, our data is not just a commodity, but a reflection of our identity, preferences, and behavior.
AI BYTE 2 📢 : The Algorithm of Thoughts (AoT): A Breakthrough Technique by Microsoft for Improving AI Reasoning
⭐ “Algorithm of Thoughts” (AoT), developed by Microsoft and Virginia Technical University is a novel approach to make large language models (LLMs) such as ChatGPT more efficient and human-like in their reasoning abilities.
AoT is inspired by both humans and machines, as it fuses the strengths of intuitive cognition and algorithmic logic.
The technique guides the LLM through a more streamlined problem-solving path, using algorithmic examples as prompts. This enables the model to explore different solutions in an organized manner, systematically balancing costs and computations.
The researchers claim that AoT outperforms previous in-context learning techniques such as the “Chain-of-Thought” (CoT) and the “Tree of Thoughts” (ToT), which can provide incorrect or inefficient intermediate steps.
They also say that AoT can rival external tree-search tools, which require more resources and time. Moreover, they suggest that AoT can improve the model’s “intuition”, as it learns to surpass the algorithm itself.
I find this approach very innovative and promising, as it represents a shift from supervised learning to integrating the search process itself.
With refinements to prompt engineering, I believe this technique can enable LLMs to solve complex real-world problems more effectively, while also reducing their carbon impact.
AI BYTE 3 📢 : How the Pentagon is Using AI to Stay Ahead of China and Russia?
⭐ One of the most important and challenging domains in AI is defense, whereit can provide a decisive edge over adversaries.
That’s why I was intrigued by the Pentagon’s new Replicator initiative, which was unveiled by Deputy Secretary of Defense Kathleen Hicks at a Defense Innovation Conference.
The goal, she said, is to “field Attritable Autonomous Systems at scale of multiple thousands, in multiple domains, within the next 18 to 24 months.”
“Attritable” means the robots are cheap enough to be “placed at risk and lost if the mission is of high priority,” Hicks explained—basically the military equivalent of pawns on a chess board.
The goal is rapidly developing swarms of smart, disposable air, land, and sea vehicles that can outmaneuver enemies.
Hicks cited China’s growing military as the key motivation, saying the U.S. will counter their “mass with mass of our own” using American ingenuity.
Right now only a handful of developed countries are using and exploring AI for military operations. According to an AI arms race compilation on Wikipedia, the list includes China, Russia, Israel, and some EU members.
The U.S. is so far highlighting defensive, not offensive, applications. According to an official announcement, this includes an AI-enabled airspace monitoring system scheduled to be deployed around Washington D.C.
Using advanced computer vision technology built by “non-traditional Defense Department vendor” Teleidoscope, it promises to enhance threat detection capabilities compared to existing 9/11-era systems.
The announcement says the upgrade delivers “a tenfold increase in performance capability” in identifying suspicious aircraft.
“ [The system leverages] market advancements in machine learning and augmented reality features in surveillance cameras that assist air battle managers with their ability to identify flying objects within NCR airspace," Lt.Col Kurtis Engleson said in the announcement.
Teleidoscope’s technology underwent 18 months of testing before being awarded a $100 million production contract this month. The automation and improved response times will save time, money, and lives, according to Heidi Shyu, Undersecretary of Defense for Research and Engineering.
“We are able to rapidly identify operational needs and materialize them into usable national defense solutions. This saves time and money, but more importantly, the decision advantage gained by technologies like this will save lives," she said.
This surge in AI investment by the military is intended to empower the U.S. military with leading-edge capabilities while maintaining what Hicks described as "our responsible and ethical approach to AI and autonomous systems."
With competing nations like China and Russia racing to deploy ever-more advanced AI, the technology gap appears to be closing quickly. For the Pentagon, innovating ahead of these threats is becoming an urgent priority, and they are already innovating.
The U.S. military is already training AI to handle State secrets, and even working on AI fighter jets, and the technology has yielded good—albeit early—results.