Facebook bets on open source to compete in Artificial Intelligence

Facebook spreads its artificial intelligence technology in the open.

Repeating the play that worked so well for Google with Android and Chrome, Facebook bets on open source to compete in Artificial Intelligence. In this sense, it is following the opposite path to that of the other big players: OpenAI (Microsoft), Google and Baidu, who consider the decision dangerous.

As I said above, the decision to open source is not new, although it does not always achieve the expected results. Netscape opened the source of its browser to compete with Internet Explorer and Sun that of its office suite. Although Firefox knew how to be relatively successful in the browser market, it is losing more and more ground to Chrome, and neither OpenOfiice and its fork LibreOffice have achieved a significant share of the market.

Facebook bets on open source

In February 2023, Meta, the parent company of Facebook, Instagram and WhatsApp, made its AI-based chatbot technology known as LlaMA available to academics, government researchers and others approved by the company. The name is short for Meta Large Language Model.

A large language model consists of systems that generate their responses through information obtained from large amounts of text.. The best known example is ChatGPT. They do identifying patterns in the text to build the responses that users make in natural language.

Although other open source models exist, Meta went further by releasing LlaMA already "trained". Although running a model has hardware resources that are relatively accessible to anyone, training requires the expensive uptime of a huge amount of highly specialized hardware.

Critics

Meta's decision drew criticism from the press and competitors. Following the current paradigm of ignoring the scientific method and choosing the preferred conclusion and discarding anything that might disprove it, researchers at Stanford University used it to generate "problem texts" such as instructions for disposing of a dead body or a defense of Adolf Hitler's views.

One of them told his colleagues that making this technology available to the general public was like:

…a grenade available to everyone in a grocery store.

Of course, it is not something that an Austrian corporal uses it to write a book called "Mein Kamp" and ends up causing the second war or that a group of Italian immigrants learn to get rid of their rivals by putting cement shoes on them.

With a little more common sense than an academic opposing the free dissemination of knowledge, from Meta they argue that:

You can't stop people from creating meaningless or dangerous information or whatever. But you can prevent it from spreading.

Not that I agree. There is no need to prevent it from spreading. You have to let it spread and refute it. Or, teach why it is dangerous.

I am an adult and I already had a father, I do not need the State, nor the press, nor a university nor the big technological companies to take his place.

Google's position is that:

We want to think more carefully about revealing details or the source code of Artificial Intelligence projects. Can that lead to misuse?

Can you imagine our ancestors doing the same with the wheel, fire or the steam engine?

Of course, you have to read between the lines. LMeta's decision is based on the growing mistrust among politicians and analysts about possible violations of the privacy of users of this technology.. Let's remember that ChatGPT was banned in Italy and Google's Bard cannot be used in the European Union.

And, in the case of Google, I have my serious suspicions that the refusal to reveal its code has much more to do with the fact that it is far behind its competitors, not only in generating a usable alternative but also in making it profitable.
But, only time will tell how this goes.

Fe de erratas

Where it says Facebook it should say Meta.


Be the first to comment

Leave a Comment

Your email address will not be published. Required fields are marked with *

*

*

  1. Responsible for the data: AB Internet Networks 2008 SL
  2. Purpose of the data: Control SPAM, comment management.
  3. Legitimation: Your consent
  4. Communication of the data: The data will not be communicated to third parties except by legal obligation.
  5. Data storage: Database hosted by Occentus Networks (EU)
  6. Rights: At any time you can limit, recover and delete your information.