Investors are pouring hundreds of billions of dollars into the AI industry right now, and much of that is going toward the development of a still theoretical technology: artificial general intelligence.

OpenAI, the maker of the buzzy chatbot ChatGPT, has made creating AGI a top priority. Its Big Tech competitors, Google, Meta, and Microsoft, are also devoting their top researchers to the same goal.

AI experts have said it would likely be 2050 before AGI hits the market. OpenAI CEO Sam Altman says 2025, but it’s a very difficult problem to solve.

Let’s also emphasise ethical considerations. The Dartmouth participants didn’t spend much time discussing the ethical implications of AI. Today, we know better, and must do better.

We must also refocus research directions. Let’s emphasise research into AI interpretability and robustness, interdisciplinary AI research and explore new paradigms of intelligence that aren’t modelled on human cognition.

Finally, we must manage our expectations about AI. Sure, we can be excited about its potential. But we must also have realistic expectations, so that we can avoid the disappointment cycles of the past.

Robotic forearm designed with human-like proportions and efficient heat dissipation:

To replicate this in robots, researchers developed a compact forearm with a radioulnar joint using miniature bone–muscle modules. The design mimics human proportions, with two modules in the radius and ulna, totaling eight muscles. These muscles control six degrees of freedom (DOFs), including the radioulnar joint, radiocarpal joint, and finger movements.

The module’s compact design maintains the correct body proportions and weight ratios while offering more muscle-driven freedom than other robots. The researchers successfully created a forearm that closely mirrors human joint performance, allowing for precise, skillful movements similar to those of a human.

Researchers tested Kengoro, a robot equipped with a human-mimetic radioulnar forearm, by performing tasks like soldering, opening a book, turning a screw, and swinging a badminton racket.

The report suggests that Broadcom or Marvell will design the chip, but Apple might be a partner, too.

Last week, OpenAI published the GPT-4o “scorecard,” a report that details “key areas of risk” for the company’s latest large language model, and how they hope to mitigate them.

In one terrifying instance, OpenAI found that the model’s Advanced Voice Mode — which allows users to speak with ChatGPT — unexpectedly imitated users’ voices without their permission, Ars Technica reports.

“Voice generation can also occur in non-adversarial situations, such as our use of that ability to generate voices for ChatGPT’s advanced voice mode,” OpenAI wrote in its documentation. “During testing, we also observed rare instances where the model would unintentionally generate an output emulating the user’s voice.”

Amazon taps Anthropic’s Claude AI for a new and improved version of Alexa.