2022 started with crypto and ended with the crypto cataclysm. As an epitaph we can put the declaration of Fabio Panetta, the Italian of the board of the ECB: "Cryptocurrency is just a gambling den where the gambling is disguised as investment”. Paul Krugman immediately echoed him by stating that he had always said so. He had always really said it: apart from criminal activities, there is nothing that regulated finance cannot do, and do it better than cryptocurrencies.
It won't be a final goodbye to crypto, because the skeleton of this technology, i.e. the blockchain, will survive, but probably part ways with finance or rather that kind of crypto-finance which has reproduced, making it worse, the traditional model of intermediated and opaque transactions of which the aborted cryptocurrency exchange platform FTX it was the hypostasis.
It is therefore no surprise that the forecast for the year 2003 of the financial newspaper "Financial Times" is precisely "Farewell crypto”, with a surprising, however, “hello generative AI”.
2023 will be precisely the year of theGenerative artificial intelligence, for the London newspaper.
Towards what?
He writes in the January 3, 2023 editorial:
“Investors have fallen in love with the next big thing in the tech sector. It is said that this year will be the breakthrough year for artificial intelligence. While this statement could have been made for any of the past years, there is now a real belief that we are indeed at a turning point”.
And it could really be the good year. A nice surprise came in the last glimpse of 2022. A surprising, albeit immature, essay of potential of artificial intelligence it has left the insiders' rooms to enter the field of action of millions of people of the most disparate professions and aptitudes.
OpenAI
It was the OpenAI non-profit consortium which aims to promote actions for the benign use of AI and which it sees among its founders Elon Musk, known or well-known to all, and Sam Altman, former president of the mythical business accelerator YCombinator.
In November 2022 OpenAI launched and made one available for free chatbot named ChatGBT, a sort of language generation engine based on unsupervised machine learning technologies.
In simple words ChatGBT is able to write, program, produce information, images, audio, video from natural language inputs like those designed to launch a Google search, but, if desired, more complex and structured.
The more structured they are, the more articulated and punctual the answer is.
The Turing test
What ChatGBT returns following the type and configuration of the query can be an article, a short essay, a visual composition or lines of programming source code. But how does he produce similar things that presuppose encyclopaedic and even specialist knowledge.
Il generative engine uses an immense amount of information that it receives from the Internet and on which specialized software performs an intervention selection and amalgamation starting from the data found in his "well" of knowledge. The results can be surprising to such an extent that it really seems to be able to pass the Turing Test, in the sense of no longer being able to distinguish between a thinking machine and a biological intelligence.
This is the feeling that millions of users have had and whose experience has led many observers to think that 2023 could be the turning point for artificial intelligence.
The “Financial Times” writes that more than 160 start-up have already embarked on exploring this model to produce more advanced implementations and solutions.
The practical fallout
Generative AI has the enormous potential to increase productivity and the quality of the work produced by the creative industries. Those who work in these sectors will be able to make use of ideas and advice offered by generative AI without the excessive danger, for the moment, of being replaced in their roles by a non-biological entity.
Just as machine tools, first, and mechanical robots, later, led to unimaginable goals in industrial production, so generative AI can lead to unthinkable outlets in cognitive revolution.
It is these possible outlets, already concretely foreseeable, that tickle the instinct of investors and capital managers. It may be that the money bomb that has hitherto fallen into cryptocurrency territory spills over into this new tech region.
From an individual tool to an industrial one
For now the generative model it can go to support copywriters, those who develop software, those who have to write a film or those who attend training courses, in short, all those who are low on fuel and inspiration. In the short term however it could have a significant impact on sectors as diverse as i services to customers, the marketing , sale, la dealer it's the same Research.
Undoubtedly the entire research and services sector could be turned upside down by this technology which is not only the evolution of the existing one, but a radical reconfiguration of it.
Towards a distributed intelligence
Google, which first launched generative AI and which has the most advanced technologies kept in-house, this new wave brings a quasi-existential challenge to its current model not only technological, but also and above all commercial. It is no coincidence that the search engine is very guarded and cautious about implementing generative AI in its own services.
The problem is again who is in control of what. Who is in the button room? OpenAI's action of making such a powerful tool available to the general public is the expression of a broader trend that aims to remove the control of artificial intelligence technologies from the hands of experts and large groups to put them in those of the wider audience. This is the profound meaning of Web3, if it ever comes.
This "democratization” (inverted commas given the story), as the editorial of the “Financial Times” underlines, can have enormous implications and create extraordinary opportunities for many subjects.
For example the success that the platforms are having low code/no code software” is paving the way for an ever-widening audience to develop applications that will now be able to do without the technology teams and organizations that now tend to impose their agenda.
The nature of generative AI
Every opportunity carries risk, just as every power carries responsibility. Often i results of generative AI are wrong, incomplete or unreliable hidden in the aura of amazement that these tools generate. In reality, these models can produce discordant answers to a question set with terms that do not coincide or with different nuances.
Generative AI it is not a deterministic technology. It is not a pocket calculator, which always produces the same output, in space and time, from an arithmetic or logical input.
Generative AI is pretty a probabilistic technology which can only give a statistically probable approximation to a question. Timnit Gebru, an AI scientist and ex-Google researcher, has called applications of generative AI “stochastic parrots” (stochastic parrots).
Just because of the stochastic-parrot nature of the generative AI model, it is necessary for users to assume responsibility for verifying the truthfulness of the result before using it in any informational or professional context. Paradoxically, the task falling on the user is greater and more demanding the more sophisticated the means with which he finds himself interacting.
As the “Financial Times” writes, the product of generative AI cannot be never the last word if anything the first. Difficult to think otherwise, even if the London newspaper is often wrong, as happens to ChatBot. The intelligence and seriousness of the recipient is the fundamental element of everything.
...
From: A breakout year for artificial intelligence, “The Financial Times”, January 3, 2023
