Toggle light / dark theme

In today’s AI news, Mercor, the AI recruiting startup founded by three 21-year-old Thiel Fellows, has raised $100 million in a Series B round, the company confirmed to TechCrunch. Menlo Park-based Felicis led the round, valuing Mercor at $2 billion — eight times its previous valuation. Existing investors Benchmark and General Catalyst, as well as DST Global and Menlo Ventures participated.

In other advancements, GPT-4.5 could arrive as soon as next week, as Microsoft gets ready to host OpenAI’s latest artificial intelligence models.

Microsoft engineers are currently readying server capacity for OpenAI’s upcoming GPT-4.5 and GPT-5 models. While OpenAI CEO Sam Altman acknowledged recently that GPT-4.5 will launch within a matter of weeks.

Then, OpenAI’s astounding growth rate potential is luring possible investors as questions loom over whether the startup will go public. “In terms of a multiple to pay for stock like ours, there’s incredible interest at the moment,” finance chief Sarah Friar told CNBC’s David Faber on Thursday. Its future growth potential has also enabled OpenAI to “achieve valuations that are on par with the growth rate of the scale” it is reaching.

S internal testing, it could mark a meaningful step forward for an all-purpose multimodal AI that can operate interactively in both real and digital spaces. + In videos, Figure is introducing Helix, a generalist Vision-Language-Action (VLA) model that unifies perception, language understanding, and learned control to overcome multiple longstanding challenges in robotics. A detailed report on Helix can be found in text accompanying the video.

Then, in this episode of Moonshots Peter Diamandis is joined by a panel of leaders in the session Transforming Business with AI: Opportunity or Overload? at Miami FII. Panelists include: Prem Akkaraju, CEO, Stability AI Ramin Hasani, Co-Founder & CEO, Liquid AI Jack Hidary, CEO, SandboxAQ Jim Keller, CEO, Tenstorrent Alexander Sukharevsky, Senior Partner & Managing Partner, QuantumBlack, AI, McKinsey & Company.

Meanwhile, AI is evolving into a mysterious new form of intelligence — powerful yet flawed, capable of remarkable feats but still far from human-like reasoning and efficiency. To truly understand it and unlock its potential, we need a new science of intelligence that combines neuroscience, AI and physics, says neuroscientist and Stanford professor Surya Ganguli.

Quantum computers, which operate leveraging quantum mechanics phenomena, could eventually tackle some optimization and computational problems faster and more efficiently than their classical counterparts. Instead of bits, the fundamental units of information in classical computers, quantum computers rely on qubits (quantum bits), which can be in multiple states at once.

Silicon-based quantum dots, semiconductor-based structures that trap individual electrons, have been widely used as qubits, as the spin state of the electrons they confine can be leveraged to encode information. Despite their promise, many quantum computers developed so far are susceptible to decoherence, which entails the disruption of qubit states due to their interaction with the surrounding environment.

Researchers at the University of Rochester recently set out to experimentally realize a so-called nuclear-spin dark state, a condition that has been theorized to improve the performance of quantum computers, suppressing undesirable interactions and thus reducing decoherence. Their paper, published in Nature Physics, demonstrates the potential of this state for reducing decoherence in and thus potentially improving control over quantum information processing.

In a leap forward for quantum computing, a Microsoft team led by UC Santa Barbara physicists on Wednesday unveiled an eight-qubit topological quantum processor, the first of its kind. The chip, built as a proof-of-concept for the scientists’ design, opens the door to the development of the long-awaited topological quantum computer.

“We’ve got a bunch of stuff that we’ve been keeping under wraps that we’re dropping all at once now,” said Microsoft Station Q Director Chetan Nayak, a professor of physics at UCSB and a Technical Fellow for Quantum Hardware at Microsoft. The chip was revealed at Station Q’s annual conference in Santa Barbara, and accompanies a paper published in the journal Nature, authored by Station Q, their Microsoft teammates and a host of collaborators that presents the research team’s measurements of these new qubits.

“We have created a new state of matter called a topological superconductor,” Nayak explained. This phase of matter hosts exotic boundaries called Majorana zero modes (MZM) that are useful for , he explained. Results of rigorous simulation and testing of their heterostructure devices are consistent with the observation of such states. “It shows that we can do it, do it fast and do it accurately,” he said.

Joint research demonstrating the ability to readout superconducting qubits with an optical transducer was published in Nature Physics.

Quantum computing has the potential to drive transformative breakthroughs in fields such as advanced material design, artificial intelligence, and drug discovery. Of the quantum computing modalities, superconducting qubits are a leading platform towards realizing a practical quantum computer given their fast gate speeds and ability to leverage existing semiconductor industry manufacturing techniques.

However, fault-tolerant quantum computing will likely require 10,000 to a million physical qubits. The sheer amount of wiring, amplifiers and microwave components required to operate such large numbers of qubits far exceeds the capacity of modern-day dilution refrigerators, a core component of a superconducting quantum computing system, in terms of both space and passive heat load.

In a groundbreaking study, scientists developed new ways to control atom collisions using optical tweezers, offering insights that could advance quantum computing and molecular science. By manipulating light frequencies and atomic energy levels, they mapped out how specific atomic characteristics influence collision outcomes, paving the way for more precise quantum manipulation.

A collaborative team of researchers from the Max Planck Institute for Structure and Dynamics of Matter (MPSD), Nanjing University, Songshan Lake Materials Laboratory (SLAB), and international partners has introduced a new method to regulate exotic electronic states in two-dimensional materials.

Building on the foundations laid by their previous work on twisted van der Waals materials, the team of physicists has now discovered a novel way to manipulate correlated electronic states in twisted double bilayer tungsten diselenide (TDB-WSe₂). This breakthrough offers new possibilities for developing advanced quantum materials and devices.

By precisely twisting two bilayers of WSe₂ near a 60-degree angle and applying a perpendicular electric field, the researchers have achieved control over the interaction between two distinct electronic bands, known as the K-valley and Γ-valley bands. This tuning has led to the observation of a “valley charge-transfer insulator”—an exotic state where electron movement is highly correlated, and electrical conductivity is suppressed.

As the fundamental flaw of today’s quantum computers, improving qubit stability remains the focus of much research in this field. One such stability attempt involves so-called topological quantum computing with the use of anyons, which are two-dimensional quasiparticles. Such an approach has been claimed by Microsoft in a recent paper in Nature. This comes a few years after an earlier claim by Microsoft for much the same feat, which was found to be based on faulty science and hence retracted.

The claimed creation of anyons here involves Majorana fermions, which differ from the much more typical Dirac fermions. These Majorana fermions are bound with other such fermions as a Majorana zero mode (MZM), forming anyons that are intertwined (braided) to form what are in effect logic gates. In the Nature paper the Microsoft researchers demonstrate a superconducting indium-arsenide (InAs) nanowire-based device featuring a read-out circuit (quantum dot interferometer) with the capacitance of one of the quantum dots said to vary in a way that suggests that the nanowire device-under-test demonstrates the presence of MZMs at either end of the wire.

Microsoft has a dedicated website to their quantum computing efforts, though it remains essential to stress that this is not a confirmation until their research is replicated by independent researchers. If confirmed, MZMs could provide a way to create more reliable quantum computing circuitry that does not have to lean so heavily on error correction to get any usable output. Other, competing efforts here include such things as hybrid mechanical qubits and antimony-based qubits that should be more stable owing to their eight spin configurations.

The achievement comes after the company spent nearly two decades of research in the field, but Microsoft claims that building Majorana 1 required that it create an entirely new state of matter, which it is referring to as a topological state.

Microsoft’s quantum chip employs eight topological qubits using indium arsenide, which is a semiconductor, and aluminum, which is a superconductor.

“The difficulty of developing the right materials to create the exotic particles and their associated topological state of matter is why most quantum efforts have focused on other kinds of qubits,” the company said in a blog Wednesday.

In today’s AI news, Codeium, an AI-powered coding startup, is raising a new round of funding at a $2.85 billion valuation. The round is being led by returning investor Kleiner Perkins, the people said. The new round comes just six months after Silicon Valley-based Codeium announced that it had closed a $150 million Series C at a $1.25 billion post-money valuation.

In other advancements, a couple of weeks after the initial release of Mistral’s AI assistant, Le Chat, the company told Le Parisien that it has reached one million downloads. “Go and download Le Chat, which is made by Mistral, rather than ChatGPT by OpenAI — or something else,” French president Emmanuel Macron said in a TV interview ahead of the recent AI Action Summit in Paris.

And, Google is launching a new experiment that uses AI to help people explore more career possibilities. The company announced in a blog post on Wednesday that a new “Career Dreamer” tool can find patterns between your experiences, educational background, skills, and interests to connect you with careers that might be a good fit.

Meanwhile, Forbes’ Lance Eliot analyzes a popular mantra right now. The recent AI-industry groupthink that says we merely need to increase the so-called “thinking time” of generative AI and LLMs to get better responses. AI makers are allowing users to stipulate that the AI can expend more time and effort doing various processing before displaying a generated answer.

In videos, Microsoft’s Satya Nadella sits down with Dwarkesh Patel to talk about their new Majorana Quantum chip breakthrough, plans for artificial general intelligence, topological qubits, gaming world models, and whether Microsoft Office commoditizes LLMs, or the other way around.

Then, dive into the world of Model Context Protocol and learn how to seamlessly connect AI agents to databases, APIs, and more. IBM’s Roy Derks breaks down its components, from hosts to servers, and showcases real-world applications. Gain the knowledge to revolutionize your AI projects.