AI: more risks or opportunities for the financial sector?

The world market has already exceeded 100 billion (source Statista) and the margins for development are enormous. In addition to the entrepreneur Elon Musk - who founded Open AI together with Sam Altman, a start-up built specifically to explore and expand the potential of artificial intelligence - Microsoft, Google, Meta and more recently Amazon have also joined the race for the creation of generative AI. Over the past few days, Microsoft has gained about 10% in the stock market on the back of positive results for the first quarter of the year, driven by an increasingly resilient demand deriving from the cloud computing business and an increasingly positive outlook for services related to artificial intelligence. During meetings with analysts, CEO Satya Nadella underlined investments of around 10 billion dollars in the development of artificial intelligence systems and the launch of a new chatbot on Bing.

If the big techs mentioned are the ones who have put the greatest resources on the plate, all the multinationals of the web and innovation - and not just those of Silicon Valley - are working on their chat-bot.

The reason is simple. This technology can offer various advantages, such as the collection and analysis of large quantities of data: a process which, if used on a large scale, can have revolutionary results for many strategic sectors, from infrastructure to healthcare (just think of the impact the new powerful technological brain could have in the treatment of rare diseases). Yet it's not all rosy, indeed. There are still several areas to be explored in relation to the associated risks, especially in the economic and financial sector.

“One of them is the security and transparency of the data used by artificial intelligence algorithms. The recent dispute between ChatGPT and the Italian Privacy Guarantor have highlighted the risk that these algorithms may use personal data improperly or even illegally. If governments were to decide to regulate the use of artificial intelligence in a stringent way, the companies that are investing the most in it could have to change their growth prospects and accept negative performances, as happened with Meta with regard to investments in the metaverse " , says Corrado Cominotto, head of active asset management at Banca Generali.

Furthermore, the accuracy of the information collected by artificial intelligences may have some uncertainties for those who use this technology for data collection and analysis purposes. “In fact, one might think of taking every market update or forecast as the absolute truth, based on algorithms that often misinterpret various fundamental factors and circumstances, generating market manipulation or the creation of speculative bubbles. The margin of error, in fact, is always present even for the most advanced systems, and excessive trust in the machine, not accompanied by human intervention, could lead to investment decisions that do not adhere to the desired ones. For example, an experiment conducted by a well-known Italian newspaper generated a portfolio allocation advice for an average risk profile of 20% in cryptocurrencies, clearly not appropriate, given the volatility of this asset class".

"Some of the most important banks in the world - Cominotto recalls again - are moving to verify the scope of use of these technologies, with particular caution until there is certainty about the accuracy of the analyzes they generate. JP Morgan, on the one hand is working to implement artificial intelligence, on the other hand recently restricted the use of ChatGPT for its employees, fearing, among other things for data security".

Although the instrument has a revolutionary scope if used superficially it represents more of a cross than a delight in this phase. For example, the incorrect answer of a chat bot - in finance in the grip of the artificial intelligence bubble - can lead to the loss of even thousands of dollars. This is what happened to Google which burned 100 billion of market capitalization due to an inaccurate replica of Bard, its virtual assistant, released on the day of the public presentation of the giant's innovations in the field of AI.

 

And in the financial sphere? How to reconcile the potential of technology with the risks that its use brings with it? For Banca Generali, final decisions must always be taken by a team of professionals with the task of analyzing the information collected and making informed decisions, based on years of experience in the sector.

This could be the right recipe for exploiting the use of AI also in Italy where the Privacy Guarantor, as mentioned, had blocked the ChatGPT application for smaller customers. The jurist Pasquale Stanzione has already explained that the provision against OpenAl is justified by the failure to provide users with correct information on the use of their profiles, as well as an understandable protection of minors. The ongoing negotiations with the American group should lead to a lifting of the blockade by the end of the month.

Innovation is now difficult to stop. Not only; doing so could even be harmful. According to the observatory of the Milan Polytechnic, the entire Italian artificial intelligence market is already worth half a billion and has many promising companies, moreover in a context of scientists and researchers of international level. Again according to the Polimi Observatory, in 2022 61% of large Italian companies have already launched at least one AI project, a figure that represents 10 percentage points more than five years ago. While among SMEs, 15% have at least one AI project started (in 2021 it was 6%). In Italy the companies involved operate mainly in the commercial sector, even if great progress has been made in recent times in the medical field and in infrastructures. Among the main AI players are Ammagamma, MDOtm, CyberDyne and Ynap Group.

The margins for growth, here too, are potentially boundless. And, as Ferruccio De Bortoli argues in Corriere della Sera, it would be a risk for the country to be completely cut off from a process that will inevitably mark history. How to get out of this impasse? According to what scientist Roberto Battiston told La Repubblica, our cultural approach must be different.

For example, in teaching, the ability to interrogate algorithms and the awareness that the answers are the result of the accumulation of knowledge with a strong Western and Anglo-Saxon imprint will count.

It should not be forgotten that many universities and colleges in France, India and the United States have decided to ban it, considering the precedents in which important academic exams were passed and numerous scientific papers were written with the help of OpenAI's artificial intelligence. And this also applies to other advanced AIs, with students and researchers called upon to rewrite their texts before a commission to verify effective knowledge of the subject. In some cases, the works created with the help of ChatGPT and other similar solutions have been considered invalid.

Apart from these drawbacks, some countries, such as the United States and Israel, consider the sector to be absolutely strategic; the Emirates have created a ministry for artificial intelligence and France is even supporting the local ecosystem. Italy too will have to find its path, but with due precautions. Which must be maximal in those sectors – such as banking and finance for example – which base part of their success on two pillar values ​​such as trust and confidentiality.

 

 

 

 

 

 

“L’intelligenza artificiale sta migliorando l'analisi dei dati e la previsione del mercato nella finanza. Inoltre, sta rendendo più efficienti le operazioni finanziarie e aprendo nuove opportunità come il trading algoritmico e la consulenza finanziaria personalizzata basata sull'IA”. Così ChatGPT sintetizza i cambiamenti portati dall'intelligenza artificiale all'industria finanziaria in un testo auto-prodotto.

Ma che cos’è questo cervello sovraumano che qualcuno si è spinto ad etichettare come una nuova divinità in grado di sconvolgere equilibri, ordine, logiche e schemi consolidati di questo millennio? ChatGPT è un modello di linguaggio generativo, un sistema che è in grado di rispondere a domande e produrre testi in modo autonomo e simile a un essere umano. E’ capace, per esempio, di creare un romanzo o un testo rap in rima oppure di redigere un contratto. Al momento, lo strumento è aperto a tutti gratuitamente (solo in Italia il garante della privacy ha applicato restrizioni), probabilmente al fine di addestrarlo al meglio tramite l'interazione quotidiana con milioni di utenti in centinaia di lingue.

Come nasce? A svilupparlo è stato Open Al, startup fondata da Sam Altman ed Elon Musk. La società californiana ha ricevuto un primo investimento da un miliardo di dollari da Microsoft, che ha poi considerato di decuplicarlo a 10 miliardi, assegnando ad Open AI una valutazione di 30 miliardi. Si tratta della stessa Microsoft che a fine 2022 ha rilevato il 4% del London Stock Exhange per 2 miliardi e stretto un accordo per trasferire la borsa di Londra sul cloud. Parallelamente ogni big sta investendo nell’intelligenza artificiale sviluppando un proprio chat box. Google di recente ha investito alcuni milioni in start up ritenute promettenti (300 mln sono stati destinati ad Anthropic, di cui ha rilevato il 10% e 200 mln a Cohere). Di recente però lo stesso Musk insieme a un gruppo di accademici ha chiesto una moratoria di sei mesi all’addestramento dell’intelligenza artificiale per pausa che lo strumento sfugga agli obiettivi e ai limiti imposti dai creatori e arrechi danni all’umanità. Problema di coscienza oppure battaglia di interessi?

Nessuno sa ancora rispondere. Fatto sta che l'intelligenza artificiale è in realtà il presente della finanza e di altri settori industriali. Le piattaforme che sfruttano l’IA sono in grado di proporre autonomamente un'allocazione sulla base degli obiettivi di rendimento e dei limiti di rischio stabiliti. Sistemi simili suggeriscono, per esempio, la quota di obbligazioni e l'esposizione ai mercati emergenti ottimali. Soprattutto sono in grado di adeguarsi alle richieste e alle preferenze dei singoli gestori: il portafoglio cambierà se l'investitore fisserà come requisito la presenza di titoli Eni oppure di Btp. Questa capacità di adattamento e interpretazione è ciò che differenzia sistemi come Chat Gpt dagli algoritmi sinora adoperati dalle grandi imprese nella relazione con i clienti, perlopiù capaci di risposte binari (sì-no) a domande semplici. Si pensi ai chatbot dell'app bancarie che, salvo che per problemi d soluzione elementare, rimandano sempre all'assistenza di persone fisiche. Non si devono però sopravvalutare le potenzialità dell'AL ChatGPT è (ancora) carente di logica. Il ritardo della guida autonoma dimostra poi quanto sia difficile calare la geometria algoritmica nella caotica realtà. Difficilmente, poi, l'intelligenza artificiale potrà avere l’intuito e la rete di conoscenze di cui dispone un gestore e che si rivelano spesso cruciali per gli investimenti più redditizi. L'AI non avrà probabilmente mai le cosiddette soft skills, sempre più richieste nel mondo del lavoro e della finanza. La differenza la farà la qualità dei dati che vengono quotidianamente dati in pasto ai modelli per addestrarli.