Lazy Sunday #15

UNESCO on neurotechnology, return-to-office studies and learnings from BloombergGPT

Author

Alexander Scholz
April 21, 2024

Find my research: neuroscience

An UNESCO take on neurotechnology

For some time now, the UNESCO has brought together experts in the field of AI and neurotechnology to discuss current developments, impact and propose suggestions for collaboration and regulation. In 2023, it released a larger report as well.

Interesting takeaways:

  • The investments into the technology are increasing rapidly, not only in the deeply medical realm but also end-consumer facing: both patents, and research papers increased by 20x and 35x compared to 2000s.

  • “(…) worldwide lack of solid governance and regulation in this area is worrisome.”

  • Companies leading innovation, based on patents: IBM (US), Ping An Technology (CH), Fujitsu (JP), Microsoft (US), Samsung (KR), Sony (JP), Intel (US).

  • Neurotechnology is not just invasive BCI’, but covers sleep and performance optimization through to assistive exoskeletons.

(UNESCO)

Maximiser or Satisficers

TLDR: exercising “self-distancing” during decision-making is one of the most important tools in when facing difficult decisions. It has received continuous attention in “wise reasoning”, a fairly new field of psychology.

If you look at your problem from a third-party perspective, creating distance between your own feelings towards the problem statement, the decisions-making process becomes much more objective.

The article also dives into pretty interesting areas of decision-making in animals, who face similar dilemmas, just at a different scale. It also discussed how “maximisers”, people who can not be satisfied with the status quo, attract better outcomes but are mentally in a worse spot than “satisficers”.

This group tends to accept more eagerly, lose on 20% of potential gains (based on a study on salary negotiations), but tend to be overall happier.

Find my research: office

“Return to office!!!!”, but research suggests otherwise

THIS IS WHY I DIG INTO RESEARCH-BASED MANAGEMENT!!

It so debunks some of the old and biased thinking in management.

Many companies are demanding a sort of RTO, citing better performance and collaboration.

But productivity is not measured by when people show up.

Interestingly, the call for RTO especially in public companies, is driven often by activist investors. There is no direct correlation between financial performance and RTO mandates. If office is not the preferred place of work for the individual but rather a forced place to be, it does negatively impact the employees’ overall motivation.

Names like Atlassian have done internal research and found: remote or self-choice hybrid is king!

Offices still matter. There are people who want to be around others, need to collaborate at certain times. But when and why to collaborate and when and where to do focus work (maybe even with the new MW75 Neuro headphones) should not be mandated.

Multitasking myth

If somebody says multitasking in the office again:

First, “multitask” itself is typically a misnomer. According to experts, it’s not possible to do two things at once — unless we can do one without much thinking (like taking a walk while catching up with a friend).

“Usually, when people think they’re multitasking, they’re actually switching their attention back and forth between two separate tasks,” said Gloria Mark, a professor of informatics at the University of California, Irvine. (…)

“…we switch between tasks, we pay what’s been dubbed a ‘switch cost,’” Dr. Wagner said. “We’re going to be slower and less accurate than we would have been if we stayed on a single task.”

Entrepreneurs: a special neurodiverse group

Entrepreneurs drive to not sit still, move on, accept the biggest risks possible, and be almost blinded by their ideas is the result of a special neurological diversity, says this article. It goes so far to say most founders are neurodivergent, that is different to people who tend to thrive in a more regulated and settled environment.

While entrepreneurs account for a large part of economic growth in every economy, social security nets favour the “classically” employed in many countries. The question this article raises: can we do better to support this special group of people?

I liked the neurodivergent twist, considering we talked about neurodiversity not so long ago.

On top of tech

Tech Gadget Review: Master&Dynamics MW75 Neuro

Remember when you read your first gadget review about wearables like smartwatches? How weird was it to think that a small device could measure your heart rate and therefore arousal?

Here you go, a first review of the headphones with built-in neurotechnology.

I won’t summarize everything, you can read yourself, but it provides a good understanding of the interconnection of sensors, surrounding hardware and software.

Pretty sure, this is not the last review of use-at-home-neurotech.

AI turbo charges drug design for Parkinson by 1000x

Parkinson’s is the fastest growing neurological condition worldwide. I did not know that! Seemingly in the UK alone 1 out of 37 people will be diagnosed with Parkinson’s.

Chemical compounds in the brain, cause nerve cell death and impact many bodily functions.

Treatment is difficult as it wears off, but often it is started with high doses of dopamine and l-dopa.

Accelerating treatment, meaning reviewing the fit of chemical compounds, takes a lot of time. AI showed promising results to decreasing time by a factor 10 and cost by a factor 1000. 🤯

Another brain-inspired computing technique - Race Logic

We have talked about brain-inspired computing, such as ideas from Cortical Labs or neuromorphic chip design Rain Neuromorphics (Sam Altman).

Another idea comes from NIST, a research lab by the National Institute of Standards and Technology in the US.

The logic: instead of sending one signal after the other, parallel computing choses the problem (or solution) that arrives at a result first. Once a problem is solved, all other processes are being paused or stopped to further save energy.

This is based on how the brain computes. Neural signals aren’t stored, combined, packaged and then sent further to another area.

Instead, neurons are constantly sending and receiving information. This also means not all incoming signals are worked on.

The goal? To make computers as energy efficient as the brain.

(NIST)

$10 million for training BloombergGPT and shortly after it’s outrun

In 2023, Bloomberg trained a LLM model on its own data sets, both financial and general purpose. Bloomberg invested $10 million into the project. At that time, the trained model was outperforming existing LLMs on the market.

But when newer GPT models came out, they outperformed BloombergGPT in many areas even without the unique training set.

My guess you would receive a lift from training newer models on the same data set, but the point is current developments in AI are so rapid, that training your own models have little advantages to the next set of LLM.

The data set needs to carry over to new models. Your data is like the hardware component: it can remain the same. The LLM is the OS component, that constantly receives updates and improves your hardware.

Media Box

Ethical priorities for neurotechnologies and AI

Back in 2017, a group of people from technology companies, universities, BRAIN projects and other institutions suggested compliance and regulatory priorities for neurotechnologies. Especially the overlap of AI, personal mental information and the potential of misuse were highlighted.

What strikes me the most that regulation-wise little has changed since 2017.

Fast-forward to today, technology made significant developments. Only very few neuro-related companies existed in 2017. Neuralink was founded that year.

Since then, the technology has moved into a much more usable and wearable space, outpacing the regulatory developments.

The fear that neurotech can get into our “last resort” of personal space continues to linger till today. Even the New York Times writes about it.

Slowly, governments are moving.

Too slow, maybe.

(Nature)

Nexus Neurotech - Boutique VC’s entering the funding market for neurotech

We will see the emergence of new and small boutique venture companies focusing on neurotech. It was the same with hardware in the 1990s, internet in the 2000, SaaS in the 2010s, AI and web3 now. Nexus Neurotech is a new VC firm founded in mid-2023 by two partners who used to work at Google Life Sciences also known as Verily.

Investments are in health tech are down. Like everywhere else.

If you remove the outlier years 2020-2022, check sizes are actually consistent with 2018 and 2019 and growing. I think health had its shake up, but will increase from here, considering that AI will drive increased efficiency in research. But it’s also not SaaS - that is for sure.

Misc but not least…

In February I wrote about the Huberman visit to Sydney. Here is the QnA, especially interesting obviously from min 39:37 ;-)

Thank you for reading.

Please share the link on social media or by forwarding this episode of Lazy Sunday to your friends and colleagues.

Share the link, if you feel it is worth sharing: https://bit.ly/4d6x9Us

Maybe they can learn something as well? ;-)

Have a great rest of the weekend.

Alex

Reply

or to participate.