AI leading to a scientific revolution

[JasonPalmer]

Interesting articles in the latest edition of the Economist about how AI could speed up scientific research, automated labs and automated analysis of published material. I dont usually get the Economist but was early for a dentist appointment this morning so picked up a copy from WH Smiths, paid for it using a self service till. These machines are everywhere eh.

[Joseph K]

Great, now we just need an economic & political revolution to ensure the whole of society benefits from AI and automated work in general.

[JasonPalmer]

Great, now we just need an economic & political revolution to ensure the whole of society benefits from AI and automated work in general.

I think AI assisted journalism may help catch out politicians, we may even end up with AI assisted politicians ? Writing their speeches for the commons ?

[smittims]

I'm just relieved it's happened too late in my lifetime to affect me. Such things as faking a film of you from a sample of your voice and appearance, and the many ways AI will be used to extort money from people, fill me with dismay.

[Anastasius]

It's a shame that the forum doesn't have a Like button as Joseph and smittims would both have one from me.

[Serial_Apologist]

Great, now we just need an economic & political revolution to ensure the whole of society benefits from AI and automated work in general.

People will still be needed to manufacture commodities under capitalism, AI or no AI, because human labour is the only possible source of surplus value.

[richardfinegold]

The Media is usually behind reporting on these developments. AI has been used for years in molecular research.
It was used in the enhancement of function studies on SARS-1 in Wuhan, for example

[Bryn]

The Media is usually behind reporting on these developments. AI has been used for years in molecular research.
It was used in the enhancement of function studies on SARS-1 in Wuhan, for example

And your evidence for this? The only GoF work related to SARS-Cov-1 carried out in Wuhan (I don't know what SARS-1 is) was the 2015 US/China joint work on M15 (an avirulent, i.e. non-virulent, skeleton of SARS-Cov-1). Perhaps you know, definitively, of similar work on SARS-Cov-1 itself.

[Retune]

I don't know what role AI is supposed to have played in functional studies of SARS-CoV, the original SARS virus (often called SARS-CoV-1, though I think SARS-CoV is still the official designation). Is this a 'lab leak' claim? Various computational tools that have been used for decades in molecular biology research could be called 'AI' by the broadest definition, though they have nothing much to do with things like ChatGPT. Rather than being behind on developments, the media have in many cases been far too credulous about the 'lab leak' theory, which remains unsupported by any real evidence and is thought by most virologists (as opposed to non-specialists on Twitter) to be a much less likely explanation for the emergence of SARS-CoV-2, the COVID-19 virus, than natural spillover. Virus 'backbones' used in experiments before 2020 were either derived from SARS-CoV (MA15 is a mouse-adapted strain of SARS-CoV) or from bat viruses closely related to SARS-CoV rather than SARS-CoV-2. No credible precursor virus to SARS-CoV-2 is known to exist in any laboratory, and this was a field of research that was very open and collaborative before politically motivated 'lab leak' accusations started to be thrown around.

[richardfinegold]

The Lab Leak theory has been endorsed by the U.S. Federal Bureau of Investigation,despite a considerable amount of political pressure not to reach this conc. It’s a dogma amongst naysayers such as those here that this did not happen, and I will not spar with dogmatist. AI has routinely beeen used in facilitating genome expression and manipulation. The public became more aware of AI when its applications became used in every day consumer applications and in the arts

[Bryn]

The Lab Leak theory has been endorsed by the U.S. Federal Bureau of Investigation,despite a considerable amount of political pressure not to reach this conc. It’s a dogma amongst naysayers such as those here that this did not happen, and I will not spar with dogmatist. AI has routinely beeen used in facilitating genome expression and manipulation. The public became more aware of AI when its applications became used in every day consumer applications and in the arts

Ah, so you are not prepared to support your assertion with actual evidence, rather than citing the one US agency supporting the lab leak proposition (theory i too scientific a term to use in this context).

[Retune]

The Lab Leak theory has been endorsed by the U.S. Federal Bureau of Investigation,despite a considerable amount of political pressure not to reach this conc. It’s a dogma amongst naysayers such as those here that this did not happen, and I will not spar with dogmatist. AI has routinely beeen used in facilitating genome expression and manipulation. The public became more aware of AI when its applications became used in every day consumer applications and in the arts

If you don't consider yourself to be dogmatic, and are open to have your ideas challenged, I wonder if you'd be interested in reading the piece below? It sets out quite lucidly the ways in which the various lab leak claims have developed and changed, and the political pressures that have shaped them, not necessarily in the direction you might imagine:

https://quillette.com/2023/08/19/the-lab-leak-illusion/

(and yes, I know it's in Quillette, which isn't a publication I would normally read!).

Note that a majority of US intelligence agencies find either a natural origin most likely, or do not think they have sufficient evidence to come to a conclusion, and the recent ODNI report shoots down several long-standing lab leak arguments. The FBI opinion also appears to have been formed quite early on, and has not as far as we can tell taken on board (e.g.) recent published research demonstrating the centrality of the market in the early outbreak, and showing that plausible intermediate host animals were sold there.

If you could point to some source about the supposed role of AI in guiding manipulation of coronaviruses, that might add to the general discussion in this thread.

[JasonPalmer]

I suspect this divergence on thread topic could lead to it being deleted, do you want to create a new thread on covid origins ?

[Dave2002]

The Lab Leak theory has been endorsed by the U.S. Federal Bureau of Investigation,despite a considerable amount of political pressure not to reach this conc. It’s a dogma amongst naysayers such as those here that this did not happen, and I will not spar with dogmatist. AI has routinely beeen used in facilitating genome expression and manipulation. The public became more aware of AI when its applications became used in every day consumer applications and in the arts

Ai is certainly used in some areas of science and research, apart from itself being a subject of research. Possibly so-called Expert Systems, which were being developed and even deployed by the end of the 20th Century were based on some AI methods. Some of them were capable [seemingly] of analysing data from a patient and recommending treatments. Some of them may have needed guidance by a human, but they could produce a readable and understandable document which appeared to be sufficiently comprehensive for patient management. This would - for example - include treatments for specific diseases - but would also give alternative treatments depending on what other drugs or other treatments a patient was also having.

There are certainly also AI based image scanning processes looking for cell abnormalities which have now been shown to be more reliable than employing humans to do the same job.

Much of the so-called AI methods relies on scanning very large bodies of data, looking for patterns. As with many things in computing, the old adage GARBAGE IN - GARBAGE OUT applies.
Some of the newer "AI" approaches seem to be good, and will save time for critically important problems. Others may be poor - as they will pick up bias in the training sets, or may be deficient in other ways.

Some AI methods may be rather sophisticated and fairly reliable, and cost effective, but others may be less so. It is fairly easy to devise problems where even a very simple strategy might be effective, and then present these to an AI "learning system". If the system can be shown to be cost effective in say 90% of cases, it might be hailed as a success, whereas in fact it might have no real "understanding" of the problem domain, and be relatively useless for individual situations.

There are very significant dangers in putting one's faith in AI systems as they are at present. The problem domains matter.

[JasonPalmer]

I am looking forward to AI assisted journalism scanning hansard, government publications, statements by politicians and producing articles for newspapers.