When I was a teenager, I used to saw supercomputers as a mean to become powerful by acquiring a tool to calculate faster than competitors without relying on anyone. It now seems to me that supercomputing is less related to autonomy than it used to. Supercomputing meant autonomy by design to me. Freedom is different from autonomy: the first is spiritual while the second is material. Many hackers seem to value freedom and to pay little attention to autonomy. Eric S. Raymond praises Seymour Cray, the father of Cray supercomputers for having designed an entire system from scratch.1 This anecdote may not be entirely true but it shows that autonomy may be the best way to achieve one’s goal, for instance, to build the fastest system. Furthermore, when one has autonomously designed at least part of a solution, one can show a distinctive savoir-faire as Seymour Cray did, maybe without knowing that he was doing it.
"What is the point of having a distinctive savoir-faire that enables me to do anything on my own when I can quickly hack something to get it to do what I want?", you wonder. There are divergences regarding the importance of instincts in software programming.2 Many advocates of free or open licenses highlight their positive effect on an Anglo-Saxon conception of freedom that is political, not to say libertarian. Linus Torvalds states that kernel and interface restrictions have been crucial to facilitate portability.3 I cannot evaluate this technical decision since I do not know anything about computer science. I have relied on Contardo Ferrini’s works4 to solve practical legal issues raised by artificial intelligence although I am not a Roman lawyer. Hence, I have extremely positive feelings about borrowing and combining loosely connected legal sources to conceive custom solutions. I have been thinking of free and open licenses as an equivalent approach in software. I am still convinced that when it comes to digital, the biggest threats to autonomy are business models that foster standardisation5 and infrastructures that hinder competition6.
Nevertheless, I have realised over the past few weeks that hacking might distract from designing things well because a hack was primarily supposed to fit into a process designed by the hacker to achieve a different result. Artificial intelligence paves the way to automated hacking because it autonomously processes data to produce other pieces of data for a different purpose. The table that has been used to produce the picture of a table that has been imported into a database to become part of a dataset has little to do with the piece of furniture at the end of the process. When one paints the table, the reproduction bears some resemblance with the original and the painting is not the output of a process in contrast to artificial intelligence. Computer science does not necessarily involve hacking. Seymour Cray has designed Cray supercomputers well in the sense that their original purpose was to calculate faster than any other machine; he has achieved speed by design, not by hack. I do not know whether such an exploit would still be possible today since processing through huge infrastructures operated according to strict standardised practice seems to be the new paradigm of the digital industry.
Does the fact that a result is achieved by design and not by hack really matter?
Any artificial Intelligence has to fit into a process to be efficient. The process based on artificial intelligence may be more important than your work in the sense that you may have to integrate your work within a new process to preserve its efficiency. Artificial intelligence may also be designed to do things that a human cannot do such as revealing a hardly noticeable pattern in an image to help a practitioner to identify the symptom of a disease.
It is worth considering the following questions to see that artificial intelligence may affect your autonomy at work:
- What makes the product of your work, be it your pieces of code, you press releases, or your legal documents7 different from a particular output generated by artificial intelligence and selected by a human from a range of outputs?
The issue is not whether an artificial intelligence does a thing as you do it but whether a process can be implemented to achieve the same result without producing a thing as you do.
- What enables you to practice what you preach?
It is strange to preach freedom with tools such as email services based on massive data processing. They can only be operated by firms that can afford to maintain a big infrastructure. Using these services hinders smaller competitors.
- How close is the connection between your tools and your values?
Many people talk about reducing their carbon footprint and are connected to social networks all day long.
In brief, doing something as one pleases is different from being free to apply a process.
-
RAYMOND, Eric S. A Brief History of Hackerdom. In : DIBONA, Chris, OCKMAN, Sam and STONE, Mark (eds.), Open sources: voices from the open source revolution. Online. 1st ed. Beijing ; Sebastopol, CA : O’Reilly, 1999. ISBN 9781565925823, § Prologue: The Real Programmers. Available from: https://www.oreilly.com/openbook/opensources/book/. ↩
-
RAYMOND, Eric S. The Mail Must Get Through. In : The Cathedral and the Bazaar. Online. 2000. Available from: http://www.catb.org/~esr/writings/cathedral-bazaar/cathedral-bazaar/ar01s02.html ; confer STALLMAN, Richard M. The GNU Operating System and the Free Software Movement. In : footnote 1, § Scratching an Itch? ↩
-
TORVALDS, Linus. The Linux Edge. In : footnote 1, §§ Kernel Space and User Space and ff. ↩
-
If you are curious and wish to have a look at two pages, see for instance FERRINI, Contardo. “Furtum usus possessionisve”. Contributo alla dottrina del furto in diritto romano. In : ARANGIO-RUIZ, Vincenzo (ed.), Opere di Contardo Ferrini. Vol. 5. Online. Milano : Hoepli, 1930 , pp. 125 and f. Available from: https://archive.org/details/BRes0105725/page/n107/mode/2up. ↩
-
See Artificial intelligence and business models; Digital autonomy. ↩
-
See Heavy and soft. ↩