​

Comments (shown in italics) sent by Vivek Agarwal, Country Director for India at the Tony Blair Institute for Global Change. A summary of an initial discussion on how to address these very useful comments is shown under each one.

 

Vivek:

Your framing resonates strongly with several issues we’ve been grappling with - particularly the concentration of AI capabilities and the need to mobilize participation from low- and middle-income countries.

On the draft message itself, the core proposition is compelling. A couple of possible refinements to consider:

Sharpening the distinction between this effort as a public knowledge infrastructure vs downstream commercial. This will help foundations more clearly situate their non-financial role.

 

There is a distinct role for public infrastructure and for commercial applications of AI, but they can also benefit each other.

 

Commercial startups will benefit from having a public knowledge infrastructure as a starting base. This public infrastructure creates a platform for many users to leverage to create new technologies and products. In its absence a few large companies will control AI, and small startups will not be competitive. These startups are the source of a large part of technological innovation. They can be based in every country, including low- and medium- income countries.

Example of innovative startups in AI are DeepMind and Nvidia among many other science-based companies that were started by small teams. Given the low barrier for implementing software applications assisted by AI tools, even teams from low- and middle-income countries could develop commercial solutions, and participating in building an open knowledge infrastructure will contribute to their training.

 

Many companies already share their AI research as open-source code or as publications, to gain support from society and to attract the best talent. Both company scientists and academic scientists wish to obtain the individual recognition that comes from publication, to facilitate future career progression in multiple directions. The scientific community will encourage this trend by sharing ideas as part of the public infrastructure, making it more valuable, and by recognizing individual talent.

 

Some scientific research makes sense only as a public effort, which can have longer time horizons. Key fundamental knowledge might be too far from commercial applications to justify investments.

                   

Vivek:

Briefly clarifying how scientific credit and priority will be tracked at the individual researcher level, as this is often a key concern for institutional leaders.

 

Priority:

The date of each contribution by researchers is shown on Cellcomm.org and priority can therefore be assigned.

 

The date is also shown on the independent Internet Archive, storing the cellcomm.org pages, at

https://archive.org/

We regularly update this archive.

As the discussion grows, authors could help by actively storing on archive.org the cellcomm.org page with their contributions.

 

Future additional mirror databases could be held by participating Institutions.

 

Credit:

It is important to have multiple sources of credit, using different points of view to offer recognition. Many (possibly most) scientific advances are initially supported by a minority and using consensus will not identify them at this stage.

There is a long list of examples from the history of science supporting this observation, and many are presented on cellcomm.org.

 

The credit can be given in the future if priority if securely recorded.

 

In the process started by the draft message, institutional leaders will be contacted by Foundation and philanthropists that appreciate the ideas of their scientists.

Many Foundations have established methods for identifying promising scientific ideas.

 

Foundations will be able to see which ideas attract support from a substantial minority of scientists in the open discussion and are potential innovative approaches. The support can be explicit, or they can be shown to originate related ideas from others. The open discussion will therefore benefit the evaluation of ideas.

 

Institutional leaders can also provide recognition to their scientists for their contributions to the discussion, for example when deciding about promotions, to show their support for innovative ideas and to attract interest from foundations.

 

One of the reasons for the recent AI advances is the availability of new computer chips suitable for parallel processing, developed by companies like Nvidia. Many simple calculations are done in parallel, rather than large calculations done in sequence by a single processor. An analogy can be made with the collective processing of information by individual human scientists. In addition to major advances originating by a single scientist, which of course remain important, credit might also be given for smaller contributions, for example just reading and verifying details of the virtual models of cells and organisms, which is useful if done by many scientists.

Those that are part of the first hundreds or of the first few thousand to join the effort will be recognized for their pioneering choice to adopt new methods of open science.

​​