ODDO BHF AM

In the past, the Technology sector is well-known to have triggered a widespread climate of fever and misgivings. However, rarely has an innovation generated such enthusiasm in a short time: publicly available on November 30th, ChatGPT has gathered 1 million users in only ... 5 days! In these exceptional conditions, we couldn't miss the first public version of what is called "Generative AI" - the generation of content by an Artificial Intelligence.

What is ChatGPT?

ChatGPT is an artificial intelligence tool with a conversation channel (conversational chat), that answers questions asked by human users with speed, accuracy, and neutrality. ChatGPT has been fed and trained for three years by the company Open AI with a large amount of text from the internet. Since its public launch on November 30th, 2022, this new application has triggered an exponential interest, as seen in Alphabet Trends statistics or in the interests expressed on Twitter.

What do we know about Open AI, the company behind the launch of ChatGPT?

Open AI is an artificial intelligence research company founded in 2015 by Sam Altman and Elon Musk, among others. It aims at promoting an artificial intelligence that serves humanity. Open AI has transformed itself since its creation, going from a non-profit organization to a company that allows itself (limited) profits whose purpose is mainly to attract both the best data science talent and the capital to finance its development. Open AI is an unlisted company that has already: 1) several hundred employees, 2) raised more than 2 billion dollars in capital via investors such as Microsoft and AWS ( Amazon 's Internet Services subsidiary); 3) a trilogy of applications with - in appearance, strange names but which are destined to become standards: ChatGPT; DALL- E 2 and Whisper.

What are the first applications of ChatGPT?

The language processing model used by this artificial intelligence is known as GPT-3. It constitutes a technological revolution because it can answer to a wide range of applications. Indeed, in theory, ChatGPT provides users with answers to their questions. In practice, any everyday problem can be submitted: from the elaboration of a diet to the resolution of a mathematical problem or the production of an essay on any given theme, etc. Thus, GPT-3 clearly represents a major quantum leap compared to the previous stage of artificial intelligence, which corresponds to the Alphabet search.

Given the wide range of domains that ChatGPT can address, the number of potential users of this tool seems huge. It can range from a student looking for a quick help to write a paper (or even a thesis) to a journalist, editor or writer who wants to speed up his or her creative process, to a housewife confronted with a specific practical problem.

Three questions, among many others, that this new generation of artificial intelligence raises ...

A new competition for Alphabet in the "search" sector, its core business?

The exponential – and even “viral”, development of ChatGPT, as well as the technological breakthrough represented by its natural language-based interface model, appear to be the beginning of a real challenge to Alphabet . The blow may seem harsh for Alphabet , as its search tool ( Alphabet Search) is the top entry point to the internet. The considerable advertising revenues that Alphabet generates through this channel may be threatened. Far from denying the technological disruption that ChatGPT represents, the magnitude of the impact for Alphabet is still unclear for two reasons. 1) Ultimately, ChatGPT and Alphabet search have different purposes: the former is an augmented version of the latter. This could allow a certain coexistence between the two tools, with Alphabet Search addressing simpler user needs. 2) Alphabet is not idle and has made colossal investments in Artificial Intelligence/Machine Learning. It has considerable financial firepower and a gigantic pool of very high-level engineers to counterattack this technological breakthrough.

How to distinguish a text written by a human from one written by an AI?

As ChatGPT, or other similar conversational chats, are improved to become more accurate and efficient, it will become more and more difficult to distinguish the results produced by humans from those produced by an AI: texts, articles, etc. This anticipation, which seems to be a dystopian vision of the future, may lead to profound upheavals in many professions. While some industries could experience job creation, others could be threatened by such tools. After the workers that have had to cope with the emergence of robots in factories for the last 30 years, it may soon be the white-collar workers (lawyers, accountants, consultants, teachers, etc.) who will have to adapt to this new situation.

Facilitated malware creation: new challenges for cyber security?

Tools such as ChatGPT should contribute to making it even easier to write lines of computer code. However, hackers could be tempted to hijack these tools to create their evil malware faster, therefore, deviating from the proper usage of the tool. Fortunately, by a welcome pendulum effect, there is no doubt that artificial intelligence will be put to work to respond to these new forms of electronic attacks. As we have already discussed in previous newsflashes on this subject, we believe that cybersecurity represents a particularly vast and promising field of application for AI. In other words, cybersecurity is one of the many areas (along with health, education, etc.) where AI can offer a solution for the future, rather than raising a problem.