Without a doubt, ChatGPT is one of the most heavily discussed technological innovations of recent months. But what is actually behind Artificial Intelligence (AI) technology, and how are the big technology companies reacting to this new trend?
What is ChatGPT?
ChatGPT is an AI-based language model created by the company OpenAI. The concept and the underlying technology have been known for years, but only modern cloud computing solutions provide the necessary computing power for the very extensive and complex language models such as the one behind ChatGPT.
The abilities of ChatGPT are definitely impressive, and the response on the Internet to the release in November 2022 was accordingly massive.
Within 5 days, more than one million users had registered for the beta test phase – by comparison, Instagram needed 2.5 months, Facebook 10 months, and Netflix 3.5 years for these numbers.
User euphoria and curiosity remain unbroken even three months after the launch. At this point, the programme has more than 100 million active users, who send around 10 million search queries a day – a number that even Google Search only reached after about 2.5 years.
This comparison with traditional industries is even more striking:
- ChatGPT: 100 million users after 3 months
- Aircraft: 50 million users after 64 years
- Automobile: 50 million users after 62 years
- TV: 50 million users after 22 years
How does ChatGPT work?
The only purpose of a language model is to understand a random text input by a user and to create a suitable text response.
The text input is analysed in terms of the relation of the words to each other as well as the overall context, and ChatGPT asks itself, simply put, the following question:
“Based on the text I already know, what word should be next in this context?”
Here, the language model draws on the billions of text passages (books, Wikipedia, internet) it has been trained on and assigns words probabilities of being the next appropriate word.
This process is repeated recursively until – in an ideal scenario – a factually correct and coherent text is output.
As an example: The cat lies on …
- the sofa
- the keyboard
- the roof of a car
…and
- sleeps
- purrs
- stretches
By the way, when we tried it out, ChatGPT decided to go for “The cat lies on the sofa, sleeping peacefully.”
Will ChatGPT soon be doing my job?
Even though AI is making remarkable progress, fundamental weaknesses remain.
The technology is perfectly suited to assist with existing tasks (e.g. summarising texts, reproducing factual knowledge, etc.). However, it will not be able to make more complex decisions autonomously in the foreseeable future. A low level of logical understanding or the lack of “common sense” are reasons for this, among other issues.
For example, this could be a typical office use case: a language model sifts through hundreds of pages of scanned contracts for subject-specific relevant data (e.g. contract clauses, balance sheet ratios, etc.) and creates a summary of decision-relevant information.
How can I participate in this development?
Complex language models such as ChatGPT basically require three things:
- Know-how
- Computing power (to train the model)
- Data (to train the model)
Microsoft
Microsoft benefits from the strategic alliance with OpenAI, the company behind ChatGPT. According to media reports, Microsoft signed a multi-year contract worth around USD 10bn a few weeks ago. ChatGPT technology will be used in Office365 products and Teams, in addition to its integration with the search engine “Bing” (market share ~3%). This offers Microsoft great opportunities to offer premium features for the more than 300 million Office customers worldwide.
Google Search (market share ~90%) plans to make its own chatbot, “Bard”, available to the public in due course. Similar to the Microsoft search engine Bing, the answers of the bot will be integrated into the search engine’s results page in a separate window.
Assuming that the operation and performance of both chatbots will not differ significantly for users, I would not expect any significant shifts in search engine market shares. Google has the immense advantage of having access to a plethora of different user data due to its extremely broad product range (Google Search, Chrome Browser, Android operating system, etc.). This data, in anonymised form, is essential for training models based on AI – the higher the data quality, the better the result of the model.
Nvidia
Nvidia, a producer of graphic processing units (GPUs), benefits from the increased demand for computational power in consequence of the popularity of chatbots like ChatGPT and from the drastic long-term increase in demand for applications and solutions in AI.
Industry analysts estimate that the computational resources needed to train complex language models such as ChatGPT with new data (e.g. from the Internet) could increase 200-fold every two years.
Conclusion
Three months after its launch, ChatGPT still enjoys great popularity as the sustainably high user numbers show. Even if AI still displays some fundamental flaws, the technology could become even more important in the future. Numerous companies in the tech industry have already recognised this and are investing in applications and solutions in AI.
By investing in the equity fund ERSTE FUTURE INVEST, retail investors can also participate in this development. Currently the portfolio is invested in each of the three aforementioned companies, i.e. Microsoft, Google, and Nvidia. The US software giant Microsoft is currently top-ranked in terms of assets under management.
ERSTE FUTURE INVEST focuses on mega trends, as a result of which the fund covers a wide area of import future-related topics; among them healthcare, technology, lifestyle, and economic transition.
Benefits for investors
- Participation in global, future-oriented themes (mega trends)
- Investment in actively managed portfolio consisting of quality and growth shares
- Diversification of risk across several countries and sectors
Risks to bear in mind
- The assets held by the portfolio may be subject to price fluctuations
- Foreign exchange risk may affect the price of the fund
- Capital loss is possible
- The following risks may be of particular relevance to the fund: credit risk, counterparty risk, liquidity risk, deposit risk, derivative risk, and operational risks. For comprehensive information on the risks of the fund, please refer to the prospectus and to the information for investors according to sec. 21, part II, chapter “Risk notices” of the Austrian Alternative Investment Fund Managers Act
For a glossary of technical terms, please visit this link: Fund Glossary | Erste Asset Management
Legal note:
Prognoses are no reliable indicator for future performance.
For a glossary of technical terms, please visit this link: Fund Glossary | Erste Asset Management
Please note that an investment in securities also entails risks in addition to the opportunities described.