Seidor
señor usando un ordenador

March 07, 2023

ChatGPT, Google and... Microsoft. Different battle, same actors

For some time now, we have been talking about ChatGPT and the Generative AI - that is, AI built around algorithms capable of creating new content of all kinds thanks to machine learning - that gives rise to it. At the time, almost everyone was amazed by the possibilities exhibited by this sort of toy that, in some way, challenged many of our notions as casual users of technology.

What is ChatGPT?

Although it is hard to find someone in our industry who hasn't heard of it, let's recall that ChatGPT is a free chatbot (although it has an associated pay version) that can generate a response to almost any question. Developed by the company OpenAI atop a model known as GPT 3.5, or Generative Pre-Trained Transformer, and presented to the public in November 2022, it is currently considered a leader of sorts among present and past AI automatons.

In addition to answering everyday questions, it can help develop a story - even the plot of a novel, do children's homework, and more. It can even write computer code. In general, it should be able to create almost any type of content, although the platform also provides the means to detect if a text was written with help from AI; in other words, it can identify cheating (https://platform.openai.com/ai-text-classifier).

As one might expect, some have managed to deceive both the chatbot and the aforementioned AI detection tool. They even got the chatbot to reveal confidential information, such as some of its internal operating rules (this was the case of student Kevin Liu, who integrated ChatGPT into Microsoft's Bing search engine).

Yes, search engines. The appearance of ChatGPT posed a practical and immediate challenge to the status quo of traditional search platforms - essentially the all-powerful Google - which were caught woefully off-guard by the smart capabilities of chatbots to give intelligent and articulate answers on any topic (by relying, logically, on all the existing search platforms).

So the gauntlet was thrown, beyond the initial astonishment we referred to at the beginning. The traditional competition between search engines, the undisputed winner of which had been, for many years, the giant with the multicolour letter logo, was shifting, spurred on by a seemingly insignificant company by comparison, to searches with intelligent content. And there, friends, things aren't so clear at the moment.

Google's response to ChatGPT's success - and, above all, to calling into serious question its hegemony in the field of searches and content - didn't take long (or maybe it did for many). DeepMind, the AI division of the multinational from California, has been working for some time on Sparrow, its own AI chatbot (also based on the GPT model, specifically GPT 2 and GPT 3), but until then it had not gone beyond the scientific experiment category, and it was subject to all the precautions inherent to the commercial generalisation of generative AI. When this ingenious invention (presented in January as a quick response to the success of ChatGPT) was expected to be made available to the public over the course of 2023, the company suddenly changed tack and shifted the focus, just as hastily, to another seemingly different project (called Bard) as the ultimate weapon against the OpenAI chatbot.

First revolutionary invention

So what can we say about the first revolutionary invention? We spoke before about ChatGPT together with Bing, that is, Microsoft. Well, the global battle for public-use AI is boiling down to some familiar names. Despite having gone, practically overnight and without realising it, from a (mismatched) competition between search engines to an all-out battle to create smart content, the key players are basically the same old contenders.

Microsoft - which had expanded its partnership with OpenAI through a multi-billion-dollar investment in infrastructure - recently launched an update to integrate Bing with the ChatGPT engine into the Windows 11 taskbar; an implementation (currently on the to-do list) that has not been without problems, including disturbing, incorrect or downright rude answers to some users when the chatbot was challenged with certain conversations (for example, about feelings), or the aforementioned vulnerabilities that can reveal certain private information about its operating rules.

The field of AI-generated content (which will end up being, and is in fact already, a commodity) will be, without a doubt, beset by multiple threatening or, at least, disturbing aspects (manipulations, systemic biases, censorship, concentration of power in the laboratories that create the models, etc.), although the most likely outcome is that the inherent competitive dynamic (which is drawing in other giants, like Amazon with Hugging Face) will naturally end up establishing some self-regulating patterns, even if we don't yet know if they will be for the better (experience gives us reason to doubt).

You may be interested in

March 10, 2023

Corporate Venturing, a form of innovation for companies

Innovating is not always easy, especially when companies already have a certain entity, size and/or trajectory. So it is always said that small and startup companies tend to be more agile, even if they have fewer resources.

So one way to bridge the needs and opportunities of both is to reach some kind of agreement whereby the smaller one continues to lead innovation, but the larger one supports it and benefits from it.

SEIDOR
March 09, 2023

What is a warehouse management system or WMS?

If you have a business in which, either through manufacturing or distribution, you have several warehouses and store various objects in them, you know the complexities and difficulties associated with this. It is not just about having a good inventory of each and every one of the repositories and the different product lines,

SEIDOR
February 28, 2023

Data and analytics conversations

In our series of conversations and reflections on how to provide value to our customers with Analytics solutions, today we are talking with Lluis Aspachs. While everyone else was worried about the Y2K bug, Lluis was helping customers from different sectors and sizes with SAP BusinessObjects solutions across Europe and the Middle East

SEIDOR